Search results for: neural substrates
51 Assessment of DNA Sequence Encoding Techniques for Machine Learning Algorithms Using a Universal Bacterial Marker
Authors: Diego Santibañez Oyarce, Fernanda Bravo Cornejo, Camilo Cerda Sarabia, Belén Díaz Díaz, Esteban Gómez Terán, Hugo Osses Prado, Raúl Caulier-Cisterna, Jorge Vergara-Quezada, Ana Moya-Beltrán
Abstract:
The advent of high-throughput sequencing technologies has revolutionized genomics, generating vast amounts of genetic data that challenge traditional bioinformatics methods. Machine learning addresses these challenges by leveraging computational power to identify patterns and extract information from large datasets. However, biological sequence data, being symbolic and non-numeric, must be converted into numerical formats for machine learning algorithms to process effectively. So far, some encoding methods, such as one-hot encoding or k-mers, have been explored. This work proposes additional approaches for encoding DNA sequences in order to compare them with existing techniques and determine if they can provide improvements or if current methods offer superior results. Data from the 16S rRNA gene, a universal marker, was used to analyze eight bacterial groups that are significant in the pulmonary environment and have clinical implications. The bacterial genes included in this analysis are Prevotella, Abiotrophia, Acidovorax, Streptococcus, Neisseria, Veillonella, Mycobacterium, and Megasphaera. These data were downloaded from the NCBI database in Genbank file format, followed by a syntactic analysis to selectively extract relevant information from each file. For data encoding, a sequence normalization process was carried out as the first step. From approximately 22,000 initial data points, a subset was generated for testing purposes. Specifically, 55 sequences from each bacterial group met the length criteria, resulting in an initial sample of approximately 440 sequences. The sequences were encoded using different methods, including one-hot encoding, k-mers, Fourier transform, and Wavelet transform. Various machine learning algorithms, such as support vector machines, random forests, and neural networks, were trained to evaluate these encoding methods. The performance of these models was assessed using multiple metrics, including the confusion matrix, ROC curve, and F1 Score, providing a comprehensive evaluation of their classification capabilities. The results show that accuracies between encoding methods vary by up to approximately 15%, with the Fourier transform obtaining the best results for the evaluated machine learning algorithms. These findings, supported by the detailed analysis using the confusion matrix, ROC curve, and F1 Score, provide valuable insights into the effectiveness of different encoding methods and machine learning algorithms for genomic data analysis, potentially improving the accuracy and efficiency of bacterial classification and related genomic studies.Keywords: DNA encoding, machine learning, Fourier transform, Fourier transformation
Procedia PDF Downloads 1150 Tip60’s Novel RNA-Binding Function Modulates Alternative Splicing of Pre-mRNA Targets Implicated in Alzheimer’s Disease
Authors: Felice Elefant, Akanksha Bhatnaghar, Keegan Krick, Elizabeth Heller
Abstract:
Context: The severity of Alzheimer’s Disease (AD) progression involves an interplay of genetics, age, and environmental factors orchestrated by histone acetyltransferase (HAT) mediated neuroepigenetic mechanisms. While disruption of Tip60 HAT action in neural gene control is implicated in AD, alternative mechanisms underlying Tip60 function remain unexplored. Altered RNA splicing has recently been highlighted as a widespread hallmark in the AD transcriptome that is implicated in the disease. Research Aim: The aim of this study was to identify a novel RNA binding/splicing function for Tip60 in human hippocampus and impaired in brains from AD fly models and AD patients. Methodology/Analysis: The authors used RNA immunoprecipitation using RNA isolated from 200 pooled wild type Drosophila brains for each of the 3 biological replicates. To identify Tip60’s RNA targets, they performed genome sequencing (DNB-SequencingTM technology, BGI genomics) on 3 replicates for Input RNA and RNA IPs by Tip60. Findings: The authors' transcriptomic analysis of RNA bound to Tip60 by Tip60-RNA immunoprecipitation (RIP) revealed Tip60 RNA targets enriched for critical neuronal processes implicated in AD. Remarkably, 79% of Tip60’s RNA targets overlap with its chromatin gene targets, supporting a model by which Tip60 orchestrates bi-level transcriptional regulation at both the chromatin and RNA level, a function unprecedented for any HAT to date. Since RNA splicing occurs co-transcriptionally and splicing defects are implicated in AD, the authors investigated whether Tip60-RNA targeting modulates splicing decisions and if this function is altered in AD. Replicate multivariate analysis of transcript splicing (rMATS) analysis of RNA-Seq data sets from wild-type and AD fly brains revealed a multitude of mammalian-like AS defects. Strikingly, over half of these altered RNAs were bonafide Tip60-RNA targets enriched for in the AD-gene curated database, with some AS alterations prevented against by increasing Tip60 in fly brain. Importantly, human orthologs of several Tip60-modulated spliced genes in Drosophila are well characterized aberrantly spliced genes in human AD brains, implicating disruption of Tip60’s splicing function in AD pathogenesis. Theoretical Importance: The authors' findings support a novel RNA interaction and splicing regulatory function for Tip60 that may underlie AS impairments that hallmark AD etiology. Data Collection: The authors collected data from RNA immunoprecipitation experiments using RNA isolated from 200 pooled wild type Drosophila brains for each of the 3 biological replicates. They also performed genome sequencing (DNBSequencingTM technology, BGI genomics) on 3 replicates for Input RNA and RNA IPs by Tip60. Questions: The question addressed by this study was whether Tip60 has a novel RNA binding/splicing function in human hippocampus and whether this function is impaired in brains from AD fly models and AD patients. Conclusions: The authors' findings support a novel RNA interaction and splicing regulatory function for Tip60 that may underlie AS impairments that hallmark AD etiology.Keywords: Alzheimer's disease, cognition, aging, neuroepigenetics
Procedia PDF Downloads 6849 Modeling Visual Memorability Assessment with Autoencoders Reveals Characteristics of Memorable Images
Authors: Elham Bagheri, Yalda Mohsenzadeh
Abstract:
Image memorability refers to the phenomenon where certain images are more likely to be remembered by humans than others. It is a quantifiable and intrinsic attribute of an image. Understanding how visual perception and memory interact is important in both cognitive science and artificial intelligence. It reveals the complex processes that support human cognition and helps to improve machine learning algorithms by mimicking the brain's efficient data processing and storage mechanisms. To explore the computational underpinnings of image memorability, this study examines the relationship between an image's reconstruction error, distinctiveness in latent space, and its memorability score. A trained autoencoder is used to replicate human-like memorability assessment inspired by the visual memory game employed in memorability estimations. This study leverages a VGG-based autoencoder that is pre-trained on the vast ImageNet dataset, enabling it to recognize patterns and features that are common to a wide and diverse range of images. An empirical analysis is conducted using the MemCat dataset, which includes 10,000 images from five broad categories: animals, sports, food, landscapes, and vehicles, along with their corresponding memorability scores. The memorability score assigned to each image represents the probability of that image being remembered by participants after a single exposure. The autoencoder is finetuned for one epoch with a batch size of one, attempting to create a scenario similar to human memorability experiments where memorability is quantified by the likelihood of an image being remembered after being seen only once. The reconstruction error, which is quantified as the difference between the original and reconstructed images, serves as a measure of how well the autoencoder has learned to represent the data. The reconstruction error of each image, the error reduction, and its distinctiveness in latent space are calculated and correlated with the memorability score. Distinctiveness is measured as the Euclidean distance between each image's latent representation and its nearest neighbor within the autoencoder's latent space. Different structural and perceptual loss functions are considered to quantify the reconstruction error. The results indicate that there is a strong correlation between the reconstruction error and the distinctiveness of images and their memorability scores. This suggests that images with more unique distinct features that challenge the autoencoder's compressive capacities are inherently more memorable. There is also a negative correlation between the reduction in reconstruction error compared to the autoencoder pre-trained on ImageNet, which suggests that highly memorable images are harder to reconstruct, probably due to having features that are more difficult to learn by the autoencoder. These insights suggest a new pathway for evaluating image memorability, which could potentially impact industries reliant on visual content and mark a step forward in merging the fields of artificial intelligence and cognitive science. The current research opens avenues for utilizing neural representations as instruments for understanding and predicting visual memory.Keywords: autoencoder, computational vision, image memorability, image reconstruction, memory retention, reconstruction error, visual perception
Procedia PDF Downloads 7548 Employing Remotely Sensed Soil and Vegetation Indices and Predicting by Long Short-Term Memory to Irrigation Scheduling Analysis
Authors: Elham Koohikerade, Silvio Jose Gumiere
Abstract:
In this research, irrigation is highlighted as crucial for improving both the yield and quality of potatoes due to their high sensitivity to soil moisture changes. The study presents a hybrid Long Short-Term Memory (LSTM) model aimed at optimizing irrigation scheduling in potato fields in Quebec City, Canada. This model integrates model-based and satellite-derived datasets to simulate soil moisture content, addressing the limitations of field data. Developed under the guidance of the Food and Agriculture Organization (FAO), the simulation approach compensates for the lack of direct soil sensor data, enhancing the LSTM model's predictions. The model was calibrated using indices like Surface Soil Moisture (SSM), Normalized Vegetation Difference Index (NDVI), Enhanced Vegetation Index (EVI), and Normalized Multi-band Drought Index (NMDI) to effectively forecast soil moisture reductions. Understanding soil moisture and plant development is crucial for assessing drought conditions and determining irrigation needs. This study validated the spectral characteristics of vegetation and soil using ECMWF Reanalysis v5 (ERA5) and Moderate Resolution Imaging Spectrometer (MODIS) data from 2019 to 2023, collected from agricultural areas in Dolbeau and Peribonka, Quebec. Parameters such as surface volumetric soil moisture (0-7 cm), NDVI, EVI, and NMDI were extracted from these images. A regional four-year dataset of soil and vegetation moisture was developed using a machine learning approach combining model-based and satellite-based datasets. The LSTM model predicts soil moisture dynamics hourly across different locations and times, with its accuracy verified through cross-validation and comparison with existing soil moisture datasets. The model effectively captures temporal dynamics, making it valuable for applications requiring soil moisture monitoring over time, such as anomaly detection and memory analysis. By identifying typical peak soil moisture values and observing distribution shapes, irrigation can be scheduled to maintain soil moisture within Volumetric Soil Moisture (VSM) values of 0.25 to 0.30 m²/m², avoiding under and over-watering. The strong correlations between parcels suggest that a uniform irrigation strategy might be effective across multiple parcels, with adjustments based on specific parcel characteristics and historical data trends. The application of the LSTM model to predict soil moisture and vegetation indices yielded mixed results. While the model effectively captures the central tendency and temporal dynamics of soil moisture, it struggles with accurately predicting EVI, NDVI, and NMDI.Keywords: irrigation scheduling, LSTM neural network, remotely sensed indices, soil and vegetation monitoring
Procedia PDF Downloads 3647 Developmental Difficulties Prevalence and Management Capacities among Children Including Genetic Disease in a North Coastal District of Andhra Pradesh, India: A Cross-sectional Study
Authors: Koteswara Rao Pagolu, Raghava Rao Tamanam
Abstract:
The present study was aimed to find out the prevalence of DD's in Visakhapatnam, one of the north coastal districts of Andhra Pradesh, India during a span of five years. A cross-sectional investigation was held at District early intervention center (DEIC), Visakhapatnam from 2016 to 2020. To identify the pattern and trend of different DD's including seasonal variations, a retrospective analysis of the health center's inpatient database for the past 5 years was done. Male and female children aged 2 months-18 years are included in the study with the prior permission of the concerned medical officer. The screening tool developed by the Ministry of health and family welfare, India, was used for the study. Among 26,423 cases, children with birth defects are 962, 2229 with deficiencies, 7516 with diseases, and 15716 with disabilities were admitted during the study period. From birth defects, congenital deafness occurred in large numbers with 22.66%, and neural tube defect observed in a small number of cases with 0.83% during the period. From the side of deficiencies, severe acute malnutrition has mostly occurred (66.80 %) and a small number of children were affected with goiter (1.70%). Among the diseases, dental carriers (67.97%) are mostly found and these cases were at peak during the years 2016 and 2019. From disabilities, children with vision impairment (20.55%) have mostly approached the center. Over the past 5 years, the admission rate of down's syndrome and congenital deafness cases showed a rising trend up to 2019 and then declined. Hearing impairment, motor delay, and learning disorder showed a steep rise and gradual decline trend, whereas severe anemia, vitamin-D deficiency, otitis media, reactive airway disease, and attention deficit hyperactivity disorder showed a declining trend. However, congenital heart diseases, dental caries, and vision impairment admission rates showed a zigzag pattern over the past 5 years. This center had inadequate diagnostic facilities related to genetic disease management. For advanced confirmation, the cases are referred to a district government hospital or private diagnostic laboratories in the city for genetic tests. Information regarding the overall burden and pattern of admissions in the health center is obtained by the review of DEIC records. Through this study, it is observed that the incidence of birth defects, as well as genetic disease burden, is high in the Visakhapatnam district. Hence there is a need for strengthening of management services for these diseases in this region.Keywords: child health screening, developmental delays, district early intervention center, genetic disease management, infrastructural facility, Visakhapatnam district
Procedia PDF Downloads 20246 Self-Organizing Maps for Exploration of Partially Observed Data and Imputation of Missing Values in the Context of the Manufacture of Aircraft Engines
Authors: Sara Rejeb, Catherine Duveau, Tabea Rebafka
Abstract:
To monitor the production process of turbofan aircraft engines, multiple measurements of various geometrical parameters are systematically recorded on manufactured parts. Engine parts are subject to extremely high standards as they can impact the performance of the engine. Therefore, it is essential to analyze these databases to better understand the influence of the different parameters on the engine's performance. Self-organizing maps are unsupervised neural networks which achieve two tasks simultaneously: they visualize high-dimensional data by projection onto a 2-dimensional map and provide clustering of the data. This technique has become very popular for data exploration since it provides easily interpretable results and a meaningful global view of the data. As such, self-organizing maps are usually applied to aircraft engine condition monitoring. As databases in this field are huge and complex, they naturally contain multiple missing entries for various reasons. The classical Kohonen algorithm to compute self-organizing maps is conceived for complete data only. A naive approach to deal with partially observed data consists in deleting items or variables with missing entries. However, this requires a sufficient number of complete individuals to be fairly representative of the population; otherwise, deletion leads to a considerable loss of information. Moreover, deletion can also induce bias in the analysis results. Alternatively, one can first apply a common imputation method to create a complete dataset and then apply the Kohonen algorithm. However, the choice of the imputation method may have a strong impact on the resulting self-organizing map. Our approach is to address simultaneously the two problems of computing a self-organizing map and imputing missing values, as these tasks are not independent. In this work, we propose an extension of self-organizing maps for partially observed data, referred to as missSOM. First, we introduce a criterion to be optimized, that aims at defining simultaneously the best self-organizing map and the best imputations for the missing entries. As such, missSOM is also an imputation method for missing values. To minimize the criterion, we propose an iterative algorithm that alternates the learning of a self-organizing map and the imputation of missing values. Moreover, we develop an accelerated version of the algorithm by entwining the iterations of the Kohonen algorithm with the updates of the imputed values. This method is efficiently implemented in R and will soon be released on CRAN. Compared to the standard Kohonen algorithm, it does not come with any additional cost in terms of computing time. Numerical experiments illustrate that missSOM performs well in terms of both clustering and imputation compared to the state of the art. In particular, it turns out that missSOM is robust to the missingness mechanism, which is in contrast to many imputation methods that are appropriate for only a single mechanism. This is an important property of missSOM as, in practice, the missingness mechanism is often unknown. An application to measurements on one type of part is also provided and shows the practical interest of missSOM.Keywords: imputation method of missing data, partially observed data, robustness to missingness mechanism, self-organizing maps
Procedia PDF Downloads 14545 Seismic Perimeter Surveillance System (Virtual Fence) for Threat Detection and Characterization Using Multiple ML Based Trained Models in Weighted Ensemble Voting
Authors: Vivek Mahadev, Manoj Kumar, Neelu Mathur, Brahm Dutt Pandey
Abstract:
Perimeter guarding and protection of critical installations require prompt intrusion detection and assessment to take effective countermeasures. Currently, visual and electronic surveillance are the primary methods used for perimeter guarding. These methods can be costly and complicated, requiring careful planning according to the location and terrain. Moreover, these methods often struggle to detect stealthy and camouflaged insurgents. The object of the present work is to devise a surveillance technique using seismic sensors that overcomes the limitations of existing systems. The aim is to improve intrusion detection, assessment, and characterization by utilizing seismic sensors. Most of the similar systems have only two types of intrusion detection capability viz., human or vehicle. In our work we could even categorize further to identify types of intrusion activity such as walking, running, group walking, fence jumping, tunnel digging and vehicular movements. A virtual fence of 60 meters at GCNEP, Bahadurgarh, Haryana, India, was created by installing four underground geophones at a distance of 15 meters each. The signals received from these geophones are then processed to find unique seismic signatures called features. Various feature optimization and selection methodologies, such as LightGBM, Boruta, Random Forest, Logistics, Recursive Feature Elimination, Chi-2 and Pearson Ratio were used to identify the best features for training the machine learning models. The trained models were developed using algorithms such as supervised support vector machine (SVM) classifier, kNN, Decision Tree, Logistic Regression, Naïve Bayes, and Artificial Neural Networks. These models were then used to predict the category of events, employing weighted ensemble voting to analyze and combine their results. The models were trained with 1940 training events and results were evaluated with 831 test events. It was observed that using the weighted ensemble voting increased the efficiency of predictions. In this study we successfully developed and deployed the virtual fence using geophones. Since these sensors are passive, do not radiate any energy and are installed underground, it is impossible for intruders to locate and nullify them. Their flexibility, quick and easy installation, low costs, hidden deployment and unattended surveillance make such systems especially suitable for critical installations and remote facilities with difficult terrain. This work demonstrates the potential of utilizing seismic sensors for creating better perimeter guarding and protection systems using multiple machine learning models in weighted ensemble voting. In this study the virtual fence achieved an intruder detection efficiency of over 97%.Keywords: geophone, seismic perimeter surveillance, machine learning, weighted ensemble method
Procedia PDF Downloads 7444 ExactData Smart Tool For Marketing Analysis
Authors: Aleksandra Jonas, Aleksandra Gronowska, Maciej Ścigacz, Szymon Jadczak
Abstract:
Exact Data is a smart tool which helps with meaningful marketing content creation. It helps marketers achieve this by analyzing the text of an advertisement before and after its publication on social media sites like Facebook or Instagram. In our research we focus on four areas of natural language processing (NLP): grammar correction, sentiment analysis, irony detection and advertisement interpretation. Our research has identified a considerable lack of NLP tools for the Polish language, which specifically aid online marketers. In light of this, our research team has set out to create a robust and versatile NLP tool for the Polish language. The primary objective of our research is to develop a tool that can perform a range of language processing tasks in this language, such as sentiment analysis, text classification, text correction and text interpretation. Our team has been working diligently to create a tool that is accurate, reliable, and adaptable to the specific linguistic features of Polish, and that can provide valuable insights for a wide range of marketers needs. In addition to the Polish language version, we are also developing an English version of the tool, which will enable us to expand the reach and impact of our research to a wider audience. Another area of focus in our research involves tackling the challenge of the limited availability of linguistically diverse corpora for non-English languages, which presents a significant barrier in the development of NLP applications. One approach we have been pursuing is the translation of existing English corpora, which would enable us to use the wealth of linguistic resources available in English for other languages. Furthermore, we are looking into other methods, such as gathering language samples from social media platforms. By analyzing the language used in social media posts, we can collect a wide range of data that reflects the unique linguistic characteristics of specific regions and communities, which can then be used to enhance the accuracy and performance of NLP algorithms for non-English languages. In doing so, we hope to broaden the scope and capabilities of NLP applications. Our research focuses on several key NLP techniques including sentiment analysis, text classification, text interpretation and text correction. To ensure that we can achieve the best possible performance for these techniques, we are evaluating and comparing different approaches and strategies for implementing them. We are exploring a range of different methods, including transformers and convolutional neural networks (CNNs), to determine which ones are most effective for different types of NLP tasks. By analyzing the strengths and weaknesses of each approach, we can identify the most effective techniques for specific use cases, and further enhance the performance of our tool. Our research aims to create a tool, which can provide a comprehensive analysis of advertising effectiveness, allowing marketers to identify areas for improvement and optimize their advertising strategies. The results of this study suggest that a smart tool for advertisement analysis can provide valuable insights for businesses seeking to create effective advertising campaigns.Keywords: NLP, AI, IT, language, marketing, analysis
Procedia PDF Downloads 7843 The Potential Role of Some Nutrients and Drugs in Providing Protection from Neurotoxicity Induced by Aluminium in Rats
Authors: Azza A. Ali, Abeer I. Abd El-Fattah, Shaimaa S. Hussein, Hanan A. Abd El-Samea, Karema Abu-Elfotuh
Abstract:
Background: Aluminium (Al) represents an environmental risk factor. Exposure to high levels of Al causes neurotoxic effects and different diseases. Vinpocetine is widely used to improve cognitive functions, it possesses memory-protective and memory-enhancing properties and has the ability to increase cerebral blood flow and glucose uptake. Cocoa bean represents a rich source of iron as well as a potent antioxidant. It can protect from the impact of free radicals, reduces stress as well as depression and promotes better memory and concentration. Wheatgrass is primarily used as a concentrated source of nutrients. It contains vitamins, minerals, carbohydrates, amino acids and possesses antioxidant and anti-inflammatory activities. Coenzyme Q10 (CoQ10) is an intracellular antioxidant and mitochondrial membrane stabilizer. It is effective in improving cognitive disorders and has been used as anti-aging. Zinc is a structural element of many proteins and signaling messenger that is released by neural activity at many central excitatory synapses. Objective: To study the role of some nutrients and drugs as Vinpocetine, Cocoa, Wheatgrass, CoQ10 and Zinc against neurotoxicity induced by Al in rats as well as to compare between their potency in providing protection. Methods: Seven groups of rats were used and received daily for three weeks AlCl3 (70 mg/kg, IP) for Al-toxicity model groups except for the control group which received saline. All groups of Al-toxicity model except one group (non-treated) were co-administered orally together with AlCl3 the following treatments; Vinpocetine (20mg/kg), Cocoa powder (24mg/kg), Wheat grass (100mg/kg), CoQ10 (200mg/kg) or Zinc (32mg/kg). Biochemical changes in the rat brain as acetyl cholinesterase (ACHE), Aβ, brain derived neurotrophic factor (BDNF), inflammatory mediators (TNF-α, IL-1β), oxidative parameters (MDA, SOD, TAC) were estimated for all groups besides histopathological examinations in different brain regions. Results: Neurotoxicity and neurodegenerations in the rat brain after three weeks of Al exposure were indicated by the significant increase in Aβ, ACHE, MDA, TNF-α, IL-1β, DNA fragmentation together with the significant decrease in SOD, TAC, BDNF and confirmed by the histopathological changes in the brain. On the other hand, co-administration of each of Vinpocetine, Cocoa, Wheatgrass, CoQ10 or Zinc together with AlCl3 provided protection against hazards of neurotoxicity and neurodegenerations induced by Al, their protection were indicated by the decrease in Aβ, ACHE, MDA, TNF-α, IL-1β, DNA fragmentation together with the increase in SOD, TAC, BDNF and confirmed by the histopathological examinations of different brain regions. Vinpocetine and Cocoa showed the most pronounced protection while Zinc provided the least protective effects than the other used nutrients and drugs. Conclusion: Different degrees of protection from neurotoxicity and neuronal degenerations induced by Al could be achieved through the co-administration of some nutrients and drugs during its exposure. Vinpocetine and Cocoa provided the most protection than Wheat grass, CoQ10 or Zinc which showed the least protective effects.Keywords: aluminum, neurotoxicity, vinpocetine, cocoa, wheat grass, coenzyme Q10, Zinc, rats
Procedia PDF Downloads 24442 Deep Learning Framework for Predicting Bus Travel Times with Multiple Bus Routes: A Single-Step Multi-Station Forecasting Approach
Authors: Muhammad Ahnaf Zahin, Yaw Adu-Gyamfi
Abstract:
Bus transit is a crucial component of transportation networks, especially in urban areas. Any intelligent transportation system must have accurate real-time information on bus travel times since it minimizes waiting times for passengers at different stations along a route, improves service reliability, and significantly optimizes travel patterns. Bus agencies must enhance the quality of their information service to serve their passengers better and draw in more travelers since people waiting at bus stops are frequently anxious about when the bus will arrive at their starting point and when it will reach their destination. For solving this issue, different models have been developed for predicting bus travel times recently, but most of them are focused on smaller road networks due to their relatively subpar performance in high-density urban areas on a vast network. This paper develops a deep learning-based architecture using a single-step multi-station forecasting approach to predict average bus travel times for numerous routes, stops, and trips on a large-scale network using heterogeneous bus transit data collected from the GTFS database. Over one week, data was gathered from multiple bus routes in Saint Louis, Missouri. In this study, Gated Recurrent Unit (GRU) neural network was followed to predict the mean vehicle travel times for different hours of the day for multiple stations along multiple routes. Historical time steps and prediction horizon were set up to 5 and 1, respectively, which means that five hours of historical average travel time data were used to predict average travel time for the following hour. The spatial and temporal information and the historical average travel times were captured from the dataset for model input parameters. As adjacency matrices for the spatial input parameters, the station distances and sequence numbers were used, and the time of day (hour) was considered for the temporal inputs. Other inputs, including volatility information such as standard deviation and variance of journey durations, were also included in the model to make it more robust. The model's performance was evaluated based on a metric called mean absolute percentage error (MAPE). The observed prediction errors for various routes, trips, and stations remained consistent throughout the day. The results showed that the developed model could predict travel times more accurately during peak traffic hours, having a MAPE of around 14%, and performed less accurately during the latter part of the day. In the context of a complicated transportation network in high-density urban areas, the model showed its applicability for real-time travel time prediction of public transportation and ensured the high quality of the predictions generated by the model.Keywords: gated recurrent unit, mean absolute percentage error, single-step forecasting, travel time prediction.
Procedia PDF Downloads 6641 Optimization of Territorial Spatial Functional Partitioning in Coal Resource-based Cities Based on Ecosystem Service Clusters - The Case of Gujiao City in Shanxi Province
Authors: Gu Sihao
Abstract:
The coordinated development of "ecology-production-life" in cities has been highly concerned by the country, and the transformation development and sustainable development of resource-based cities have become a hot research topic at present. As an important part of China's resource-based cities, coal resource-based cities have the characteristics of large number and wide distribution. However, due to the adjustment of national energy structure and the gradual exhaustion of urban coal resources, the development vitality of coal resource-based cities is gradually reduced. In many studies, the deterioration of ecological environment in coal resource-based cities has become the main problem restricting their urban transformation and sustainable development due to the "emphasis on economy and neglect of ecology". Since the 18th National Congress of the Communist Party of China (CPC), the Central Government has been deepening territorial space planning and development. On the premise of optimizing territorial space development pattern, it has completed the demarcation of ecological protection red lines, carried out ecological zoning and ecosystem evaluation, which have become an important basis and scientific guarantee for ecological modernization and ecological civilization construction. Grasp the regional multiple ecosystem services is the precondition of the ecosystem management, and the relationship between the multiple ecosystem services study, ecosystem services cluster can identify the interactions between multiple ecosystem services, and on the basis of the characteristics of the clusters on regional ecological function zoning, to better Social-Ecological system management. Based on this cognition, this study optimizes the spatial function zoning of Gujiao, a coal resource-based city, in order to provide a new theoretical basis for its sustainable development. This study is based on the detailed analysis of characteristics and utilization of Gujiao city land space, using SOFM neural networks to identify local ecosystem service clusters, according to the cluster scope and function of ecological function zoning of space partition balance and coordination between different ecosystem services strength, establish a relationship between clusters and land use, and adjust the functions of territorial space within each zone. Then, according to the characteristics of coal resources city and national spatial function zoning characteristics, as the driving factors of land change, by cellular automata simulation program, such as simulation under different restoration strategy situation of urban future development trend, and provides relevant theories and technical methods for the "third-line" demarcations of Gujiao's territorial space planning, optimizes territorial space functions, and puts forward targeted strategies for the promotion of regional ecosystem services, providing theoretical support for the improvement of human well-being and sustainable development of resource-based cities.Keywords: coal resource-based city, territorial spatial planning, ecosystem service cluster, gmop model, geosos-FLUS model, functional zoning optimization and upgrading
Procedia PDF Downloads 5440 Hybrid Data-Driven Drilling Rate of Penetration Optimization Scheme Guided by Geological Formation and Historical Data
Authors: Ammar Alali, Mahmoud Abughaban, William Contreras Otalvora
Abstract:
Optimizing the drilling process for cost and efficiency requires the optimization of the rate of penetration (ROP). ROP is the measurement of the speed at which the wellbore is created, in units of feet per hour. It is the primary indicator of measuring drilling efficiency. Maximization of the ROP can indicate fast and cost-efficient drilling operations; however, high ROPs may induce unintended events, which may lead to nonproductive time (NPT) and higher net costs. The proposed ROP optimization solution is a hybrid, data-driven system that aims to improve the drilling process, maximize the ROP, and minimize NPT. The system consists of two phases: (1) utilizing existing geological and drilling data to train the model prior, and (2) real-time adjustments of the controllable dynamic drilling parameters [weight on bit (WOB), rotary speed (RPM), and pump flow rate (GPM)] that direct influence on the ROP. During the first phase of the system, geological and historical drilling data are aggregated. After, the top-rated wells, as a function of high instance ROP, are distinguished. Those wells are filtered based on NPT incidents, and a cross-plot is generated for the controllable dynamic drilling parameters per ROP value. Subsequently, the parameter values (WOB, GPM, RPM) are calculated as a conditioned mean based on physical distance, following Inverse Distance Weighting (IDW) interpolation methodology. The first phase is concluded by producing a model of drilling best practices from the offset wells, prioritizing the optimum ROP value. This phase is performed before the commencing of drilling. Starting with the model produced in phase one, the second phase runs an automated drill-off test, delivering live adjustments in real-time. Those adjustments are made by directing the driller to deviate two of the controllable parameters (WOB and RPM) by a small percentage (0-5%), following the Constrained Random Search (CRS) methodology. These minor incremental variations will reveal new drilling conditions, not explored before through offset wells. The data is then consolidated into a heat-map, as a function of ROP. A more optimum ROP performance is identified through the heat-map and amended in the model. The validation process involved the selection of a planned well in an onshore oil field with hundreds of offset wells. The first phase model was built by utilizing the data points from the top-performing historical wells (20 wells). The model allows drillers to enhance decision-making by leveraging existing data and blending it with live data in real-time. An empirical relationship between controllable dynamic parameters and ROP was derived using Artificial Neural Networks (ANN). The adjustments resulted in improved ROP efficiency by over 20%, translating to at least 10% saving in drilling costs. The novelty of the proposed system lays is its ability to integrate historical data, calibrate based geological formations, and run real-time global optimization through CRS. Those factors position the system to work for any newly drilled well in a developing field event.Keywords: drilling optimization, geological formations, machine learning, rate of penetration
Procedia PDF Downloads 12639 Is Brain Death Reversal Possible in Near Future: Intrathecal Sodium Nitroprusside (SNP) Superfusion in Brain Death Patients=The 10,000 Fold Effect
Authors: Vinod Kumar Tewari, Mazhar Husain, Hari Kishan Das Gupta
Abstract:
Background: Primary or secondary brain death is also accompanied with vasospasm of the perforators other than tissue disruption & further exaggerates the anoxic damage, in the form of neuropraxia. In normal conditions the excitatory impulse propagates as anterograde neurotransmission (ANT) and at the level of synapse, glutamate activates NMDA receptors on postsynaptic membrane. Nitric oxide (NO) is produced by Nitric oxide Synthetase (NOS) in postsynaptic dendride or cell body and travels backwards across a chemical synapse to bind to the axon terminal of a presynaptic neuron for regulation of ANT this process is called as the retrograde neurotransmission (RNT). Thus the primary function of NO is RNT and the purpose of RNT is regulation of chemical neurotransmission at synapse. For this reason, RNT allows neural circuits to create feedback loops. The haem is the ligand binding site of NO receptor (sGC) at presynaptic membrane. The affinity of haem exhibits > 10,000-fold excess for NO than Oxygen (THE 10,000 FOLD EFFECT). In pathological conditions ANT, normal synaptic activity including RNT is absent. NO donors like sodium nitroprusside (SNP) releases NO by activating NOS at the level of postsynaptic area. NO now travels backwards across a chemical synapse to bind to the haem of NO receptor at axon terminal of a presynaptic neuron as in normal condition. NO now acts as impulse generator (at presynaptic membrane) thus bypasses the normal ANT. Also the arteriolar perforators are having Nitric Oxide Synthetase (NOS) at the adventitial side (outer border) on which sodium nitroprusside (SNP) acts; causing release of Nitric Oxide (NO) which vasodilates the perforators causing gush of blood in brain’s tissue and reversal of brain death. Objective: In brain death cases we only think for various transplantations but this study being a pilot study reverses some criteria of brain death by vasodilating the arteriolar perforators. To study the effect of intrathecal sodium nitroprusside (IT SNP) in cases of brain death in which: 1. Retrograde transmission = assessed by the hyperacute timings of reversal 2. The arteriolar perforator vasodilatation caused by NO and the maintenance of reversal of brain death reversal. Methods: 35 year old male, who became brain death after head injury and has not shown any signs of improvement after every maneuver for 6 hours, a single superfusion done by SNP via transoptic canal route for quadrigeminal cistern and cisternal puncture for IV ventricular with SNP done. Results: He showed spontaneous respiration (7 bouts) with TCD studies showing start of pulsations of various branches of common carotid arteries. Conclusions: In future we can give this SNP via transoptic canal route and in IV ventricle before declaring the body to be utilized for transplantations or dead or in broader way we can say that in near future it is possible to revert back from brain death or we have to modify our criterion.Keywords: brain death, intrathecal sodium nitroprusside, TCD studies, perforators, vasodilatations, retrograde transmission, 10, 000 fold effect
Procedia PDF Downloads 39738 Comparison of Machine Learning-Based Models for Predicting Streptococcus pyogenes Virulence Factors and Antimicrobial Resistance
Authors: Fernanda Bravo Cornejo, Camilo Cerda Sarabia, Belén Díaz Díaz, Diego Santibañez Oyarce, Esteban Gómez Terán, Hugo Osses Prado, Raúl Caulier-Cisterna, Jorge Vergara-Quezada, Ana Moya-Beltrán
Abstract:
Streptococcus pyogenes is a gram-positive bacteria involved in a wide range of diseases and is a major-human-specific bacterial pathogen. In Chile, this year the 'Ministerio de Salud' declared an alert due to the increase in strains throughout the year. This increase can be attributed to the multitude of factors including antimicrobial resistance (AMR) and Virulence Factors (VF). Understanding these VF and AMR is crucial for developing effective strategies and improving public health responses. Moreover, experimental identification and characterization of these pathogenic mechanisms are labor-intensive and time-consuming. Therefore, new computational methods are required to provide robust techniques for accelerating this identification. Advances in Machine Learning (ML) algorithms represent the opportunity to refine and accelerate the discovery of VF associated with Streptococcus pyogenes. In this work, we evaluate the accuracy of various machine learning models in predicting the virulence factors and antimicrobial resistance of Streptococcus pyogenes, with the objective of providing new methods for identifying the pathogenic mechanisms of this organism.Our comprehensive approach involved the download of 32,798 genbank files of S. pyogenes from NCBI dataset, coupled with the incorporation of data from Virulence Factor Database (VFDB) and Antibiotic Resistance Database (CARD) which contains sequences of AMR gene sequence and resistance profiles. These datasets provided labeled examples of both virulent and non-virulent genes, enabling a robust foundation for feature extraction and model training. We employed preprocessing, characterization and feature extraction techniques on primary nucleotide/amino acid sequences and selected the optimal more for model training. The feature set was constructed using sequence-based descriptors (e.g., k-mers and One-hot encoding), and functional annotations based on database prediction. The ML models compared are logistic regression, decision trees, support vector machines, neural networks among others. The results of this work show some differences in accuracy between the algorithms, these differences allow us to identify different aspects that represent unique opportunities for a more precise and efficient characterization and identification of VF and AMR. This comparative analysis underscores the value of integrating machine learning techniques in predicting S. pyogenes virulence and AMR, offering potential pathways for more effective diagnostic and therapeutic strategies. Future work will focus on incorporating additional omics data, such as transcriptomics, and exploring advanced deep learning models to further enhance predictive capabilities.Keywords: antibiotic resistance, streptococcus pyogenes, virulence factors., machine learning
Procedia PDF Downloads 1137 Vehicle Timing Motion Detection Based on Multi-Dimensional Dynamic Detection Network
Authors: Jia Li, Xing Wei, Yuchen Hong, Yang Lu
Abstract:
Detecting vehicle behavior has always been the focus of intelligent transportation, but with the explosive growth of the number of vehicles and the complexity of the road environment, the vehicle behavior videos captured by traditional surveillance have been unable to satisfy the study of vehicle behavior. The traditional method of manually labeling vehicle behavior is too time-consuming and labor-intensive, but the existing object detection and tracking algorithms have poor practicability and low behavioral location detection rate. This paper proposes a vehicle behavior detection algorithm based on the dual-stream convolution network and the multi-dimensional video dynamic detection network. In the videos, the straight-line behavior of the vehicle will default to the background behavior. The Changing lanes, turning and turning around are set as target behaviors. The purpose of this model is to automatically mark the target behavior of the vehicle from the untrimmed videos. First, the target behavior proposals in the long video are extracted through the dual-stream convolution network. The model uses a dual-stream convolutional network to generate a one-dimensional action score waveform, and then extract segments with scores above a given threshold M into preliminary vehicle behavior proposals. Second, the preliminary proposals are pruned and identified using the multi-dimensional video dynamic detection network. Referring to the hierarchical reinforcement learning, the multi-dimensional network includes a Timer module and a Spacer module, where the Timer module mines time information in the video stream and the Spacer module extracts spatial information in the video frame. The Timer and Spacer module are implemented by Long Short-Term Memory (LSTM) and start from an all-zero hidden state. The Timer module uses the Transformer mechanism to extract timing information from the video stream and extract features by linear mapping and other methods. Finally, the model fuses time information and spatial information and obtains the location and category of the behavior through the softmax layer. This paper uses recall and precision to measure the performance of the model. Extensive experiments show that based on the dataset of this paper, the proposed model has obvious advantages compared with the existing state-of-the-art behavior detection algorithms. When the Time Intersection over Union (TIoU) threshold is 0.5, the Average-Precision (MP) reaches 36.3% (the MP of baselines is 21.5%). In summary, this paper proposes a vehicle behavior detection model based on multi-dimensional dynamic detection network. This paper introduces spatial information and temporal information to extract vehicle behaviors in long videos. Experiments show that the proposed algorithm is advanced and accurate in-vehicle timing behavior detection. In the future, the focus will be on simultaneously detecting the timing behavior of multiple vehicles in complex traffic scenes (such as a busy street) while ensuring accuracy.Keywords: vehicle behavior detection, convolutional neural network, long short-term memory, deep learning
Procedia PDF Downloads 12236 Exploring the Application of IoT Technology in Lower Limb Assistive Devices for Rehabilitation during the Golden Period of Stroke Patients with Hemiplegia
Authors: Ching-Yu Liao, Ju-Joan Wong
Abstract:
Recent years have shown a trend of younger stroke patients and an increase in ischemic strokes with the rise in stroke incidence. This has led to a growing demand for telemedicine, particularly during the COVID-19 pandemic, which has made the need for telemedicine even more urgent. This shift in healthcare is also closely related to advancements in Internet of Things (IoT) technology. Stroke-induced hemiparesis is a significant issue for patients. The medical community believes that if intervention occurs within three to six months of stroke onset, 80% of the residual effects can be restored to normal, a period known as the stroke golden period. During this time, patients undergo treatment and rehabilitation, and neural plasticity is at its best. Lower limb rehabilitation for stroke generally includes exercises such as support standing and walking posture, typically involving the healthy limb to guide the affected limb to achieve rehabilitation goals. Existing gait training aids in hospitals usually involve balance gait, sitting posture training, and precise muscle control, effectively addressing issues of poor gait, insufficient muscle activity, and inability to train independently during recovery. However, home training aids, such as braced and wheeled devices, often rely on the healthy limb to pull the affected limb, leading to lower usage of the affected limb, worsening circular walking, and compensatory movement issues. IoT technology connects devices via the internet to record, receive data, provide feedback, and adjust equipment for intelligent effects. Therefore, this study aims to explore how IoT can be integrated into existing gait training aids to monitor and sensor home rehabilitation movements, improve gait training compensatory issues through real-time feedback, and enable healthcare professionals to quickly understand patient conditions and enhance medical communication. To understand the needs of hemiparetic patients, a review of relevant literature from the past decade will be conducted. From the perspective of user experience, participant observation will be used to explore the use of home training aids by stroke patients and therapists, and interviews with physical therapists will be conducted to obtain professional opinions and practical experiences. Design specifications for home training aids for hemiparetic patients will be summarized. Applying IoT technology to lower limb training aids for stroke hemiparesis can help promote walking function recovery in hemiparetic patients, reduce muscle atrophy, and allow healthcare professionals to immediately grasp patient conditions and adjust gait training plans based on collected and analyzed information. Exploring these potential development directions provides a valuable reference for the further application of IoT technology in the field of medical rehabilitation.Keywords: stroke, hemiplegia, rehabilitation, gait training, internet of things technology
Procedia PDF Downloads 1335 High-Fidelity Materials Screening with a Multi-Fidelity Graph Neural Network and Semi-Supervised Learning
Authors: Akeel A. Shah, Tong Zhang
Abstract:
Computational approaches to learning the properties of materials are commonplace, motivated by the need to screen or design materials for a given application, e.g., semiconductors and energy storage. Experimental approaches can be both time consuming and costly. Unfortunately, computational approaches such as ab-initio electronic structure calculations and classical or ab-initio molecular dynamics are themselves can be too slow for the rapid evaluation of materials, often involving thousands to hundreds of thousands of candidates. Machine learning assisted approaches have been developed to overcome the time limitations of purely physics-based approaches. These approaches, on the other hand, require large volumes of data for training (hundreds of thousands on many standard data sets such as QM7b). This means that they are limited by how quickly such a large data set of physics-based simulations can be established. At high fidelity, such as configuration interaction, composite methods such as G4, and coupled cluster theory, gathering such a large data set can become infeasible, which can compromise the accuracy of the predictions - many applications require high accuracy, for example band structures and energy levels in semiconductor materials and the energetics of charge transfer in energy storage materials. In order to circumvent this problem, multi-fidelity approaches can be adopted, for example the Δ-ML method, which learns a high-fidelity output from a low-fidelity result such as Hartree-Fock or density functional theory (DFT). The general strategy is to learn a map between the low and high fidelity outputs, so that the high-fidelity output is obtained a simple sum of the physics-based low-fidelity and correction, Although this requires a low-fidelity calculation, it typically requires far fewer high-fidelity results to learn the correction map, and furthermore, the low-fidelity result, such as Hartree-Fock or semi-empirical ZINDO, is typically quick to obtain, For high-fidelity outputs the result can be an order of magnitude or more in speed up. In this work, a new multi-fidelity approach is developed, based on a graph convolutional network (GCN) combined with semi-supervised learning. The GCN allows for the material or molecule to be represented as a graph, which is known to improve accuracy, for example SchNet and MEGNET. The graph incorporates information regarding the numbers of, types and properties of atoms; the types of bonds; and bond angles. They key to the accuracy in multi-fidelity methods, however, is the incorporation of low-fidelity output to learn the high-fidelity equivalent, in this case by learning their difference. Semi-supervised learning is employed to allow for different numbers of low and high-fidelity training points, by using an additional GCN-based low-fidelity map to predict high fidelity outputs. It is shown on 4 different data sets that a significant (at least one order of magnitude) increase in accuracy is obtained, using one to two orders of magnitude fewer low and high fidelity training points. One of the data sets is developed in this work, pertaining to 1000 simulations of quinone molecules (up to 24 atoms) at 5 different levels of fidelity, furnishing the energy, dipole moment and HOMO/LUMO.Keywords: .materials screening, computational materials, machine learning, multi-fidelity, graph convolutional network, semi-supervised learning
Procedia PDF Downloads 2534 Fiber Stiffness Detection of GFRP Using Combined ABAQUS and Genetic Algorithms
Authors: Gyu-Dong Kim, Wuk-Jae Yoo, Sang-Youl Lee
Abstract:
Composite structures offer numerous advantages over conventional structural systems in the form of higher specific stiffness and strength, lower life-cycle costs, and benefits such as easy installation and improved safety. Recently, there has been a considerable increase in the use of composites in engineering applications and as wraps for seismic upgrading and repairs. However, these composites deteriorate with time because of outdated materials, excessive use, repetitive loading, climatic conditions, manufacturing errors, and deficiencies in inspection methods. In particular, damaged fibers in a composite result in significant degradation of structural performance. In order to reduce the failure probability of composites in service, techniques to assess the condition of the composites to prevent continual growth of fiber damage are required. Condition assessment technology and nondestructive evaluation (NDE) techniques have provided various solutions for the safety of structures by means of detecting damage or defects from static or dynamic responses induced by external loading. A variety of techniques based on detecting the changes in static or dynamic behavior of isotropic structures has been developed in the last two decades. These methods, based on analytical approaches, are limited in their capabilities in dealing with complex systems, primarily because of their limitations in handling different loading and boundary conditions. Recently, investigators have introduced direct search methods based on metaheuristics techniques and artificial intelligence, such as genetic algorithms (GA), simulated annealing (SA) methods, and neural networks (NN), and have promisingly applied these methods to the field of structural identification. Among them, GAs attract our attention because they do not require a considerable amount of data in advance in dealing with complex problems and can make a global solution search possible as opposed to classical gradient-based optimization techniques. In this study, we propose an alternative damage-detection technique that can determine the degraded stiffness distribution of vibrating laminated composites made of Glass Fiber-reinforced Polymer (GFRP). The proposed method uses a modified form of the bivariate Gaussian distribution function to detect degraded stiffness characteristics. In addition, this study presents a method to detect the fiber property variation of laminated composite plates from the micromechanical point of view. The finite element model is used to study free vibrations of laminated composite plates for fiber stiffness degradation. In order to solve the inverse problem using the combined method, this study uses only first mode shapes in a structure for the measured frequency data. In particular, this study focuses on the effect of the interaction among various parameters, such as fiber angles, layup sequences, and damage distributions, on fiber-stiffness damage detection.Keywords: stiffness detection, fiber damage, genetic algorithm, layup sequences
Procedia PDF Downloads 26233 Cicadas: A Clinician-assisted, Closed-loop Technology, Mobile App for Adolescents with Autism Spectrum Disorders
Authors: Bruno Biagianti, Angela Tseng, Kathy Wannaviroj, Allison Corlett, Megan DuBois, Kyu Lee, Suma Jacob
Abstract:
Background: ASD is characterized by pervasive Sensory Processing Abnormalities (SPA) and social cognitive deficits that persist throughout the course of the illness and have been linked to functional abnormalities in specific neural systems that underlie the perception, processing, and representation of sensory information. SPA and social cognitive deficits are associated with difficulties in interpersonal relationships, poor development of social skills, reduced social interactions and lower academic performance. Importantly, they can hamper the effects of established evidence-based psychological treatments—including PEERS (Program for the Education and Enrichment of Relationship Skills), a parent/caregiver-assisted, 16-weeks social skills intervention—which nonetheless requires a functional brain capable of assimilating and retaining information and skills. As a matter of fact, some adolescents benefit from PEERS more than others, calling for strategies to increase treatment response rates. Objective: We will present interim data on CICADAS (Care Improving Cognition for ADolescents on the Autism Spectrum)—a clinician-assisted, closed-loop technology mobile application for adolescents with ASD. Via ten mobile assessments, CICADAS captures data on sensory processing abnormalities and associated cognitive deficits. These data populate a machine learning algorithm that tailors the delivery of ten neuroplasticity-based social cognitive training (NB-SCT) exercises targeting sensory processing abnormalities. Methods: In collaboration with the Autism Spectrum and Neurodevelopmental Disorders Clinic at the University of Minnesota, we conducted a fully remote, three-arm, randomized crossover trial with adolescents with ASD to document the acceptability of CICADAS and evaluate its potential as a stand-alone treatment or as a treatment enhancer of PEERS. Twenty-four adolescents with ASD (ages 11-18) have been initially randomized to 16 weeks of PEERS + CICADAS (Arm A) vs. 16 weeks of PEERS + computer games vs. 16 weeks of CICADAS alone (Arm C). After 16 weeks, the full battery of assessments has been remotely administered. Results: We have evaluated the acceptability of CICADAS by examining adherence rates, engagement patterns, and exit survey data. We found that: 1) CICADAS is able to serve as a treatment enhancer for PEERS, inducing greater improvements in sensory processing, cognition, symptom reduction, social skills and behaviors, as well as the quality of life compared to computer games; 2) the concurrent delivery of PEERS and CICADAS induces greater improvements in study outcomes compared to CICADAS only. Conclusion: While preliminary, our results indicate that the individualized assessment and treatment approach designed in CICADAS seems effective in inducing adaptive long-term learning about social-emotional events. CICADAS-induced enhancement of processing and cognition facilitates the application of PEERS skills in the environment of adolescents with ASD, thus improving their real-world functioning.Keywords: ASD, social skills, cognitive training, mobile app
Procedia PDF Downloads 20532 Deep Learning Based Text to Image Synthesis for Accurate Facial Composites in Criminal Investigations
Authors: Zhao Gao, Eran Edirisinghe
Abstract:
The production of an accurate sketch of a suspect based on a verbal description obtained from a witness is an essential task for most criminal investigations. The criminal investigation system employs specifically trained professional artists to manually draw a facial image of the suspect according to the descriptions of an eyewitness for subsequent identification. Within the advancement of Deep Learning, Recurrent Neural Networks (RNN) have shown great promise in Natural Language Processing (NLP) tasks. Additionally, Generative Adversarial Networks (GAN) have also proven to be very effective in image generation. In this study, a trained GAN conditioned on textual features such as keywords automatically encoded from a verbal description of a human face using an RNN is used to generate photo-realistic facial images for criminal investigations. The intention of the proposed system is to map corresponding features into text generated from verbal descriptions. With this, it becomes possible to generate many reasonably accurate alternatives to which the witness can use to hopefully identify a suspect from. This reduces subjectivity in decision making both by the eyewitness and the artist while giving an opportunity for the witness to evaluate and reconsider decisions. Furthermore, the proposed approach benefits law enforcement agencies by reducing the time taken to physically draw each potential sketch, thus increasing response times and mitigating potentially malicious human intervention. With publically available 'CelebFaces Attributes Dataset' (CelebA) and additionally providing verbal description as training data, the proposed architecture is able to effectively produce facial structures from given text. Word Embeddings are learnt by applying the RNN architecture in order to perform semantic parsing, the output of which is fed into the GAN for synthesizing photo-realistic images. Rather than the grid search method, a metaheuristic search based on genetic algorithms is applied to evolve the network with the intent of achieving optimal hyperparameters in a fraction the time of a typical brute force approach. With the exception of the ‘CelebA’ training database, further novel test cases are supplied to the network for evaluation. Witness reports detailing criminals from Interpol or other law enforcement agencies are sampled on the network. Using the descriptions provided, samples are generated and compared with the ground truth images of a criminal in order to calculate the similarities. Two factors are used for performance evaluation: The Structural Similarity Index (SSIM) and the Peak Signal-to-Noise Ratio (PSNR). A high percentile output from this performance matrix should attribute to demonstrating the accuracy, in hope of proving that the proposed approach can be an effective tool for law enforcement agencies. The proposed approach to criminal facial image generation has potential to increase the ratio of criminal cases that can be ultimately resolved using eyewitness information gathering.Keywords: RNN, GAN, NLP, facial composition, criminal investigation
Procedia PDF Downloads 15231 Estimating Poverty Levels from Satellite Imagery: A Comparison of Human Readers and an Artificial Intelligence Model
Authors: Ola Hall, Ibrahim Wahab, Thorsteinn Rognvaldsson, Mattias Ohlsson
Abstract:
The subfield of poverty and welfare estimation that applies machine learning tools and methods on satellite imagery is a nascent but rapidly growing one. This is in part driven by the sustainable development goal, whose overarching principle is that no region is left behind. Among other things, this requires that welfare levels can be accurately and rapidly estimated at different spatial scales and resolutions. Conventional tools of household surveys and interviews do not suffice in this regard. While they are useful for gaining a longitudinal understanding of the welfare levels of populations, they do not offer adequate spatial coverage for the accuracy that is needed, nor are their implementation sufficiently swift to gain an accurate insight into people and places. It is this void that satellite imagery fills. Previously, this was near-impossible to implement due to the sheer volume of data that needed processing. Recent advances in machine learning, especially the deep learning subtype, such as deep neural networks, have made this a rapidly growing area of scholarship. Despite their unprecedented levels of performance, such models lack transparency and explainability and thus have seen limited downstream applications as humans generally are apprehensive of techniques that are not inherently interpretable and trustworthy. While several studies have demonstrated the superhuman performance of AI models, none has directly compared the performance of such models and human readers in the domain of poverty studies. In the present study, we directly compare the performance of human readers and a DL model using different resolutions of satellite imagery to estimate the welfare levels of demographic and health survey clusters in Tanzania, using the wealth quintile ratings from the same survey as the ground truth data. The cluster-level imagery covers all 608 cluster locations, of which 428 were classified as rural. The imagery for the human readers was sourced from the Google Maps Platform at an ultra-high resolution of 0.6m per pixel at zoom level 18, while that of the machine learning model was sourced from the comparatively lower resolution Sentinel-2 10m per pixel data for the same cluster locations. Rank correlation coefficients of between 0.31 and 0.32 achieved by the human readers were much lower when compared to those attained by the machine learning model – 0.69-0.79. This superhuman performance by the model is even more significant given that it was trained on the relatively lower 10-meter resolution satellite data while the human readers estimated welfare levels from the higher 0.6m spatial resolution data from which key markers of poverty and slums – roofing and road quality – are discernible. It is important to note, however, that the human readers did not receive any training before ratings, and had this been done, their performance might have improved. The stellar performance of the model also comes with the inevitable shortfall relating to limited transparency and explainability. The findings have significant implications for attaining the objective of the current frontier of deep learning models in this domain of scholarship – eXplainable Artificial Intelligence through a collaborative rather than a comparative framework.Keywords: poverty prediction, satellite imagery, human readers, machine learning, Tanzania
Procedia PDF Downloads 9830 Modeling Taxane-Induced Peripheral Neuropathy Ex Vivo Using Patient-Derived Neurons
Authors: G. Cunningham, E. Cantor, X. Wu, F. Shen, G. Jiang, S. Philips, C. Bales, Y. Xiao, T. R. Cummins, J. C. Fehrenbacher, B. P. Schneider
Abstract:
Background: Taxane-induced peripheral neuropathy (TIPN) is the most devastating survivorship issue for patients receiving therapy. Dose reductions due to TIPN in the curative setting lead to inferior outcomes for African American patients, as prior research has shown that this group is more susceptible to developing severe neuropathy. The mechanistic underpinnings of TIPN, however, have not been entirely elucidated. While it would be appealing to use primary tissue to study the development of TIPN, procuring nerves from patients is not realistically feasible, as nerve biopsies are painful and may result in permanent damage. Therefore, our laboratory has investigated paclitaxel-induced neuronal morphological and molecular changes using an ex vivo model of human-induced pluripotent stem cell (iPSC)-derived neurons. Methods: iPSCs are undifferentiated and endlessly dividing cells that can be generated from a patient’s somatic cells, such as peripheral blood mononuclear cells (PBMCs). We successfully reprogrammed PBMCs into iPSCs using the Erythroid Progenitor Reprograming Kit (STEMCell Technologiesᵀᴹ); pluripotency was verified by flow cytometry analysis. iPSCs were then induced into neurons using a differentiation protocol that bypasses the neural progenitor stage and uses selected small-molecule modulators of key signaling pathways (SMAD, Notch, FGFR1 inhibition, and Wnt activation). Results: Flow cytometry analysis revealed expression of core pluripotency transcription factors Nanog, Oct3/4 and Sox2 in iPSCs overlaps with commercially purchased pluripotent cell line UCSD064i-20-2. Trilineage differentiation of iPSCs was confirmed with immunofluorescent imaging with germ-layer-specific markers; Sox17 and ExoA2 for ectoderm, Nestin, and Pax6 for mesoderm, and Ncam and Brachyury for endoderm. Sensory neuron markers, β-III tubulin, and Peripherin were applied to stain the cells for the maturity of iPSC-derived neurons. Patch-clamp electrophysiology and calcitonin gene-related peptide (CGRP) release data supported the functionality of the induced neurons and provided insight into the timing for which downstream assays could be performed (week 4 post-induction). We have also performed a cell viability assay and fluorescence-activated cell sorting (FACS) using four cell-surface markers (CD184, CD44, CD15, and CD24) to select a neuronal population. At least 70% of the cells were viable in the isolated neuron population. Conclusion: We have found that these iPSC-derived neurons recapitulate mature neuronal phenotypes and demonstrate functionality. Thus, this represents a patient-derived ex vivo neuronal model to investigate the molecular mechanisms of clinical TIPN.Keywords: chemotherapy, iPSC-derived neurons, peripheral neuropathy, taxane, paclitaxel
Procedia PDF Downloads 11729 Connectomic Correlates of Cerebral Microhemorrhages in Mild Traumatic Brain Injury Victims with Neural and Cognitive Deficits
Authors: Kenneth A. Rostowsky, Alexander S. Maher, Nahian F. Chowdhury, Andrei Irimia
Abstract:
The clinical significance of cerebral microbleeds (CMBs) due to mild traumatic brain injury (mTBI) remains unclear. Here we use magnetic resonance imaging (MRI), diffusion tensor imaging (DTI) and connectomic analysis to investigate the statistical association between mTBI-related CMBs, post-TBI changes to the human connectome and neurological/cognitive deficits. This study was undertaken in agreement with US federal law (45 CFR 46) and was approved by the Institutional Review Board (IRB) of the University of Southern California (USC). Two groups, one consisting of 26 (13 females) mTBI victims and another comprising 26 (13 females) healthy control (HC) volunteers were recruited through IRB-approved procedures. The acute Glasgow Coma Scale (GCS) score was available for each mTBI victim (mean µ = 13.2; standard deviation σ = 0.4). Each HC volunteer was assigned a GCS of 15 to indicate the absence of head trauma at the time of enrollment in our study. Volunteers in the HC and mTBI groups were matched according to their sex and age (HC: µ = 67.2 years, σ = 5.62 years; mTBI: µ = 66.8 years, σ = 5.93 years). MRI [including T1- and T2-weighted volumes, gradient recalled echo (GRE)/susceptibility weighted imaging (SWI)] and gradient echo (GE) DWI volumes were acquired using the same MRI scanner type (Trio TIM, Siemens Corp.). Skull-stripping and eddy current correction were implemented. DWI volumes were processed in TrackVis (http://trackvis.org) and 3D Slicer (http://www.slicer.org). Tensors were fit to DWI data to perform DTI, and tractography streamlines were then reconstructed using deterministic tractography. A voxel classifier was used to identify image features as CMB candidates using Microbleed Anatomic Rating Scale (MARS) guidelines. For each peri-lesional DTI streamline bundle, the null hypothesis was formulated as the statement that there was no neurological or cognitive deficit associated with between-scan differences in the mean FA of DTI streamlines within each bundle. The statistical significance of each hypothesis test was calculated at the α = 0.05 level, subject to the family-wise error rate (FWER) correction for multiple comparisons. Results: In HC volunteers, the along-track analysis failed to identify statistically significant differences in the mean FA of DTI streamline bundles. In the mTBI group, significant differences in the mean FA of peri-lesional streamline bundles were found in 21 out of 26 volunteers. In those volunteers where significant differences had been found, these differences were associated with an average of ~47% of all identified CMBs (σ = 21%). In 12 out of the 21 volunteers exhibiting significant FA changes, cognitive functions (memory acquisition and retrieval, top-down control of attention, planning, judgment, cognitive aspects of decision-making) were found to have deteriorated over the six months following injury (r = -0.32, p < 0.001). Our preliminary results suggest that acute post-TBI CMBs may be associated with cognitive decline in some mTBI patients. Future research should attempt to identify mTBI patients at high risk for cognitive sequelae.Keywords: traumatic brain injury, magnetic resonance imaging, diffusion tensor imaging, connectomics
Procedia PDF Downloads 16728 “laws Drifting Off While Artificial Intelligence Thriving” – A Comparative Study with Special Reference to Computer Science and Information Technology
Authors: Amarendar Reddy Addula
Abstract:
Definition of Artificial Intelligence: Artificial intelligence is the simulation of mortal intelligence processes by machines, especially computer systems. Explicit operations of AI comprise expert systems, natural language processing, and speech recognition, and machine vision. Artificial Intelligence (AI) is an original medium for digital business, according to a new report by Gartner. The last 10 times represent an advance period in AI’s development, prodded by the confluence of factors, including the rise of big data, advancements in cipher structure, new machine literacy ways, the materialization of pall computing, and the vibrant open- source ecosystem. Influence of AI to a broader set of use cases and druggies and its gaining fashionability because it improves AI’s versatility, effectiveness, and rigidity. Edge AI will enable digital moments by employing AI for real- time analytics closer to data sources. Gartner predicts that by 2025, further than 50 of all data analysis by deep neural networks will do at the edge, over from lower than 10 in 2021. Responsible AI is a marquee term for making suitable business and ethical choices when espousing AI. It requires considering business and societal value, threat, trust, translucency, fairness, bias mitigation, explainability, responsibility, safety, sequestration, and nonsupervisory compliance. Responsible AI is ever more significant amidst growing nonsupervisory oversight, consumer prospects, and rising sustainability pretensions. Generative AI is the use of AI to induce new vestiges and produce innovative products. To date, generative AI sweats have concentrated on creating media content similar as photorealistic images of people and effects, but it can also be used for law generation, creating synthetic irregular data, and designing medicinals and accoutrements with specific parcels. AI is the subject of a wide- ranging debate in which there's a growing concern about its ethical and legal aspects. Constantly, the two are varied and nonplussed despite being different issues and areas of knowledge. The ethical debate raises two main problems the first, abstract, relates to the idea and content of ethics; the alternate, functional, and concerns its relationship with the law. Both set up models of social geste, but they're different in compass and nature. The juridical analysis is grounded on anon-formalistic scientific methodology. This means that it's essential to consider the nature and characteristics of the AI as a primary step to the description of its legal paradigm. In this regard, there are two main issues the relationship between artificial and mortal intelligence and the question of the unitary or different nature of the AI. From that theoretical and practical base, the study of the legal system is carried out by examining its foundations, the governance model, and the nonsupervisory bases. According to this analysis, throughout the work and in the conclusions, International Law is linked as the top legal frame for the regulation of AI.Keywords: artificial intelligence, ethics & human rights issues, laws, international laws
Procedia PDF Downloads 8827 Assessing P0.1 and Occlusion Pressures in Brain-Injured Patients on Pressure Support Ventilation: A Study Protocol
Authors: S. B. R. Slagmulder
Abstract:
Monitoring inspiratory effort and dynamic lung stress in patients on pressure support ventilation in the ICU is important for protecting against self inflicted lung injury (P-SILI) and diaphragm dysfunction. Strategies to address the detrimental effects of respiratory drive and effort can lead to improved patient outcomes. Two non-invasive estimation methods, occlusion pressure (Pocc) and P0.1, have been proposed for achieving lung and diaphragm protective ventilation. However, their relationship and interpretation in neuro ICU patients is not well understood. P0.1 is the airway pressure measured during a 100-millisecond occlusion of the inspiratory port. It reflects the neural drive from the respiratory centers to the diaphragm and respiratory muscles, indicating the patient's respiratory drive during the initiation of each breath. Occlusion pressure, measured during a brief inspiratory pause against a closed airway, provides information about the inspiratory muscles' strength and the system's total resistance and compliance. Research Objective: Understanding the relationship between Pocc and P0.1 in brain-injured patients can provide insights into the interpretation of these values in pressure support ventilation. This knowledge can contribute to determining extubation readiness and optimizing ventilation strategies to improve patient outcomes. The central goal is to asses a study protocol for determining the relationship between Pocc and P0.1 in brain-injured patients on pressure support ventilation and their ability to predict successful extubation. Additionally, comparing these values between brain-damaged and non-brain-damaged patients may provide valuable insights. Key Areas of Inquiry: 1. How do Pocc and P0.1 values correlate within brain injury patients undergoing pressure support ventilation? 2. To what extent can Pocc and P0.1 values serve as predictive indicators for successful extubation in patients with brain injuries? 3. What differentiates the Pocc and P0.1 values between patients with brain injuries and those without? Methodology: P0.1 and occlusion pressures are standard measurements for pressure support ventilation patients, taken by attending doctors as per protocol. We utilize electronic patient records for existing data. Unpaired T-test will be conducted to compare P0.1 and Pocc values between both study groups. Associations between P0.1 and Pocc and other study variables, such as extubation, will be explored with simple regression and correlation analysis. Depending on how the data evolve, subgroup analysis will be performed for patients with and without extubation failure. Results: While it is anticipated that neuro patients may exhibit high respiratory drive, the linkage between such elevation, quantified by P0.1, and successful extubation remains unknown The analysis will focus on determining the ability of these values to predict successful extubation and their potential impact on ventilation strategies. Conclusion: Further research is pending to fully understand the potential of these indices and their impact on mechanical ventilation in different patient populations and clinical scenarios. Understanding these relationships can aid in determining extubation readiness and tailoring ventilation strategies to improve patient outcomes in this specific patient population. Additionally, it is vital to account for the influence of sedatives, neurological scores, and BMI on respiratory drive and occlusion pressures to ensure a comprehensive analysis.Keywords: brain damage, diaphragm dysfunction, occlusion pressure, p0.1, respiratory drive
Procedia PDF Downloads 6426 CD97 and Its Role in Glioblastoma Stem Cell Self-Renewal
Authors: Niklas Ravn-Boess, Nainita Bhowmick, Takamitsu Hattori, Shohei Koide, Christopher Park, Dimitris Placantonakis
Abstract:
Background: Glioblastoma (GBM) is the most common and deadly primary brain malignancy in adults. Tumor propagation, brain invasion, and resistance to therapy critically depend on GBM stem-like cells (GSCs); however, the mechanisms that regulate GSC self-renewal are incompletely understood. Given the aggressiveness and poor prognosis of GBM, it is imperative to find biomarkers that could also translate into novel drug targets. Along these lines, we have identified a cell surface antigen, CD97 (ADGRE5), an adhesion G protein-coupled receptor (GPCR), that is expressed on GBM cells but is absent from non-neoplastic brain tissue. CD97 has been shown to promote invasiveness, angiogenesis, and migration in several human cancers, but its frequency of expression and functional role in regulating GBM growth and survival, and its potential as a therapeutic target has not been investigated. Design: We assessed CD97 mRNA and protein expression in patient derived GBM samples and cell lines using publicly available RNA-sequencing datasets and flow cytometry, respectively. To assess CD97 function, we generated shRNA lentiviral constructs that target a sequence in the CD97 extracellular domain (ECD). A scrambled shRNA (scr) with no predicted targets in the genome was used as a control. We evaluated CD97 shRNA lentivirally transduced GBM cells for Ki67, Annexin V, and DAPI. We also tested CD97 KD cells for their ability to self-renew using clonogenic tumorsphere formation assays. Further, we utilized synthetic Abs (sAbs) generated against the ECD of CD97 to test for potential antitumor effects using patient-derived GBM cell lines. Results: CD97 mRNA expression was expressed at high levels in all GBM samples available in the TCGA cohort. We found high levels of surface CD97 protein expression in 6/6 patient-derived GBM cell cultures, but not human neural stem cells. Flow cytometry confirmed downregulation of CD97 in CD97 shRNA lentivirally transduced cells. CD97 KD induced a significant reduction in cell growth in 3 independent GBM cell lines representing mesenchymal and proneural subtypes, which was accompanied by reduced (~20%) Ki67 staining and increased (~30%) apoptosis. Incubation of GBM cells with sAbs (20 ug/ ml) against the ECD of CD97 for 3 days induced GSC differentiation, as determined by the expression of GFAP and Tubulin. Using three unique GBM patient derived cultures, we found that CD97 KD attenuated the ability of GBM cells to initiate sphere formation by over 300 fold, consistent with an impairment in GSC self-renewal. Conclusion: Loss of CD97 expression in patient-derived GBM cells markedly decreases proliferation, induces cell death, and reduces tumorsphere formation. sAbs against the ECD of CD97 reduce tumorsphere formation, recapitulating the phenotype of CD97 KD, suggesting that sAbs that inhibit CD97 function exhibit anti-tumor activity. Collectively, these findings indicate that CD97 is necessary for the proliferation and survival of human GBM cells and identify CD97 as a promising therapeutically targetable vulnerability in GBM.Keywords: adhesion GPCR, CD97, GBM stem cell, glioblastoma
Procedia PDF Downloads 12825 An Integrated Lightweight Naïve Bayes Based Webpage Classification Service for Smartphone Browsers
Authors: Mayank Gupta, Siba Prasad Samal, Vasu Kakkirala
Abstract:
The internet world and its priorities have changed considerably in the last decade. Browsing on smart phones has increased manifold and is set to explode much more. Users spent considerable time browsing different websites, that gives a great deal of insight into user’s preferences. Instead of plain information classifying different aspects of browsing like Bookmarks, History, and Download Manager into useful categories would improve and enhance the user’s experience. Most of the classification solutions are server side that involves maintaining server and other heavy resources. It has security constraints and maybe misses on contextual data during classification. On device, classification solves many such problems, but the challenge is to achieve accuracy on classification with resource constraints. This on device classification can be much more useful in personalization, reducing dependency on cloud connectivity and better privacy/security. This approach provides more relevant results as compared to current standalone solutions because it uses content rendered by browser which is customized by the content provider based on user’s profile. This paper proposes a Naive Bayes based lightweight classification engine targeted for a resource constraint devices. Our solution integrates with Web Browser that in turn triggers classification algorithm. Whenever a user browses a webpage, this solution extracts DOM Tree data from the browser’s rendering engine. This DOM data is a dynamic, contextual and secure data that can’t be replicated. This proposal extracts different features of the webpage that runs on an algorithm to classify into multiple categories. Naive Bayes based engine is chosen in this solution for its inherent advantages in using limited resources compared to other classification algorithms like Support Vector Machine, Neural Networks, etc. Naive Bayes classification requires small memory footprint and less computation suitable for smartphone environment. This solution has a feature to partition the model into multiple chunks that in turn will facilitate less usage of memory instead of loading a complete model. Classification of the webpages done through integrated engine is faster, more relevant and energy efficient than other standalone on device solution. This classification engine has been tested on Samsung Z3 Tizen hardware. The Engine is integrated into Tizen Browser that uses Chromium Rendering Engine. For this solution, extensive dataset is sourced from dmoztools.net and cleaned. This cleaned dataset has 227.5K webpages which are divided into 8 generic categories ('education', 'games', 'health', 'entertainment', 'news', 'shopping', 'sports', 'travel'). Our browser integrated solution has resulted in 15% less memory usage (due to partition method) and 24% less power consumption in comparison with standalone solution. This solution considered 70% of the dataset for training the data model and the rest 30% dataset for testing. An average accuracy of ~96.3% is achieved across the above mentioned 8 categories. This engine can be further extended for suggesting Dynamic tags and using the classification for differential uses cases to enhance browsing experience.Keywords: chromium, lightweight engine, mobile computing, Naive Bayes, Tizen, web browser, webpage classification
Procedia PDF Downloads 16024 Web-Based Decision Support Systems and Intelligent Decision-Making: A Systematic Analysis
Authors: Serhat Tüzün, Tufan Demirel
Abstract:
Decision Support Systems (DSS) have been investigated by researchers and technologists for more than 35 years. This paper analyses the developments in the architecture and software of these systems, provides a systematic analysis for different Web-based DSS approaches and Intelligent Decision-making Technologies (IDT), with the suggestion for future studies. Decision Support Systems literature begins with building model-oriented DSS in the late 1960s, theory developments in the 1970s, and the implementation of financial planning systems and Group DSS in the early and mid-80s. Then it documents the origins of Executive Information Systems, online analytic processing (OLAP) and Business Intelligence. The implementation of Web-based DSS occurred in the mid-1990s. With the beginning of the new millennia, intelligence is the main focus on DSS studies. Web-based technologies are having a major impact on design, development and implementation processes for all types of DSS. Web technologies are being utilized for the development of DSS tools by leading developers of decision support technologies. Major companies are encouraging its customers to port their DSS applications, such as data mining, customer relationship management (CRM) and OLAP systems, to a web-based environment. Similarly, real-time data fed from manufacturing plants are now helping floor managers make decisions regarding production adjustment to ensure that high-quality products are produced and delivered. Web-based DSS are being employed by organizations as decision aids for employees as well as customers. A common usage of Web-based DSS has been to assist customers configure product and service according to their needs. These systems allow individual customers to design their own products by choosing from a menu of attributes, components, prices and delivery options. The Intelligent Decision-making Technologies (IDT) domain is a fast growing area of research that integrates various aspects of computer science and information systems. This includes intelligent systems, intelligent technology, intelligent agents, artificial intelligence, fuzzy logic, neural networks, machine learning, knowledge discovery, computational intelligence, data science, big data analytics, inference engines, recommender systems or engines, and a variety of related disciplines. Innovative applications that emerge using IDT often have a significant impact on decision-making processes in government, industry, business, and academia in general. This is particularly pronounced in finance, accounting, healthcare, computer networks, real-time safety monitoring and crisis response systems. Similarly, IDT is commonly used in military decision-making systems, security, marketing, stock market prediction, and robotics. Even though lots of research studies have been conducted on Decision Support Systems, a systematic analysis on the subject is still missing. Because of this necessity, this paper has been prepared to search recent articles about the DSS. The literature has been deeply reviewed and by classifying previous studies according to their preferences, taxonomy for DSS has been prepared. With the aid of the taxonomic review and the recent developments over the subject, this study aims to analyze the future trends in decision support systems.Keywords: decision support systems, intelligent decision-making, systematic analysis, taxonomic review
Procedia PDF Downloads 27323 Enhanced Multi-Scale Feature Extraction Using a DCNN by Proposing Dynamic Soft Margin SoftMax for Face Emotion Detection
Authors: Armin Nabaei, M. Omair Ahmad, M. N. S. Swamy
Abstract:
Many facial expression and emotion recognition methods in the traditional approaches of using LDA, PCA, and EBGM have been proposed. In recent years deep learning models have provided a unique platform addressing by automatically extracting the features for the detection of facial expression and emotions. However, deep networks require large training datasets to extract automatic features effectively. In this work, we propose an efficient emotion detection algorithm using face images when only small datasets are available for training. We design a deep network whose feature extraction capability is enhanced by utilizing several parallel modules between the input and output of the network, each focusing on the extraction of different types of coarse features with fined grained details to break the symmetry of produced information. In fact, we leverage long range dependencies, which is one of the main drawback of CNNs. We develop this work by introducing a Dynamic Soft-Margin SoftMax.The conventional SoftMax suffers from reaching to gold labels very soon, which take the model to over-fitting. Because it’s not able to determine adequately discriminant feature vectors for some variant class labels. We reduced the risk of over-fitting by using a dynamic shape of input tensor instead of static in SoftMax layer with specifying a desired Soft- Margin. In fact, it acts as a controller to how hard the model should work to push dissimilar embedding vectors apart. For the proposed Categorical Loss, by the objective of compacting the same class labels and separating different class labels in the normalized log domain.We select penalty for those predictions with high divergence from ground-truth labels.So, we shorten correct feature vectors and enlarge false prediction tensors, it means we assign more weights for those classes with conjunction to each other (namely, “hard labels to learn”). By doing this work, we constrain the model to generate more discriminate feature vectors for variant class labels. Finally, for the proposed optimizer, our focus is on solving weak convergence of Adam optimizer for a non-convex problem. Our noteworthy optimizer is working by an alternative updating gradient procedure with an exponential weighted moving average function for faster convergence and exploiting a weight decay method to help drastically reducing the learning rate near optima to reach the dominant local minimum. We demonstrate the superiority of our proposed work by surpassing the first rank of three widely used Facial Expression Recognition datasets with 93.30% on FER-2013, and 16% improvement compare to the first rank after 10 years, reaching to 90.73% on RAF-DB, and 100% k-fold average accuracy for CK+ dataset, and shown to provide a top performance to that provided by other networks, which require much larger training datasets.Keywords: computer vision, facial expression recognition, machine learning, algorithms, depp learning, neural networks
Procedia PDF Downloads 7222 Approximate-Based Estimation of Single Event Upset Effect on Statistic Random-Access Memory-Based Field-Programmable Gate Arrays
Authors: Mahsa Mousavi, Hamid Reza Pourshaghaghi, Mohammad Tahghighi, Henk Corporaal
Abstract:
Recently, Statistic Random-Access Memory-based (SRAM-based) Field-Programmable Gate Arrays (FPGAs) are widely used in aeronautics and space systems where high dependability is demanded and considered as a mandatory requirement. Since design’s circuit is stored in configuration memory in SRAM-based FPGAs; they are very sensitive to Single Event Upsets (SEUs). In addition, the adverse effects of SEUs on the electronics used in space are much higher than in the Earth. Thus, developing fault tolerant techniques play crucial roles for the use of SRAM-based FPGAs in space. However, fault tolerance techniques introduce additional penalties in system parameters, e.g., area, power, performance and design time. In this paper, an accurate estimation of configuration memory vulnerability to SEUs is proposed for approximate-tolerant applications. This vulnerability estimation is highly required for compromising between the overhead introduced by fault tolerance techniques and system robustness. In this paper, we study applications in which the exact final output value is not necessarily always a concern meaning that some of the SEU-induced changes in output values are negligible. We therefore define and propose Approximate-based Configuration Memory Vulnerability Factor (ACMVF) estimation to avoid overestimating configuration memory vulnerability to SEUs. In this paper, we assess the vulnerability of configuration memory by injecting SEUs in configuration memory bits and comparing the output values of a given circuit in presence of SEUs with expected correct output. In spite of conventional vulnerability factor calculation methods, which accounts any deviations from the expected value as failures, in our proposed method a threshold margin is considered depending on user-case applications. Given the proposed threshold margin in our model, a failure occurs only when the difference between the erroneous output value and the expected output value is more than this margin. The ACMVF is subsequently calculated by acquiring the ratio of failures with respect to the total number of SEU injections. In our paper, a test-bench for emulating SEUs and calculating ACMVF is implemented on Zynq-7000 FPGA platform. This system makes use of the Single Event Mitigation (SEM) IP core to inject SEUs into configuration memory bits of the target design implemented in Zynq-7000 FPGA. Experimental results for 32-bit adder show that, when 1% to 10% deviation from correct output is considered, the counted failures number is reduced 41% to 59% compared with the failures number counted by conventional vulnerability factor calculation. It means that estimation accuracy of the configuration memory vulnerability to SEUs is improved up to 58% in the case that 10% deviation is acceptable in output results. Note that less than 10% deviation in addition result is reasonably tolerable for many applications in approximate computing domain such as Convolutional Neural Network (CNN).Keywords: fault tolerance, FPGA, single event upset, approximate computing
Procedia PDF Downloads 192