Search results for: neural%20fingerprint

Authors: Liudmyla Vesper

Abstract:

The problems of modern education are large-scale and diverse. The aspirations of parents, teachers, and experts converge - everyone interested in growing up a generation of whole, well-educated persons. Both the family and society are expected in the future generation to be self-sufficient, desirable in the labor market, and capable of lifelong learning. Today's children have a powerful potential that is difficult to realize in the conditions of traditional school approaches. Focusing on STEM education in practice often ends with the simple use of computers and gadgets during class. "Science", "technology", "engineering" and "mathematics" are difficult to combine within school and university curricula, which have not changed much during the last 10 years. Solving the problems of modern education largely depends on teachers - innovators, teachers - practitioners who develop and implement effective educational methods and programs. Teachers who propose innovative pedagogical practices that allow students to master large-scale knowledge and apply it to the practical plane. Effective education considers the creation of stable neural structures during the learning process, which allow to preserve and increase knowledge throughout life. The author proposed a method of integrated lessons – cases based on the maths patterns for forming a holistic perception of the world. This method and program are scientifically substantiated and have more than 15 years of practical application experience in school and student classrooms. The first results of the practical application of the author's methodology and curriculum were announced at the International Conference "Teaching and Learning Strategies to Promote Elementary School Success", 2006, April 22-23, Yerevan, Armenia, IREX-administered 2004-2006 Multiple Component Education Project. This program is based on the concept of interdisciplinary connections and its implementation in the process of continuous learning. This allows students to save and increase knowledge throughout life according to a single pattern. The pattern principle stores information on different subjects according to one scheme (pattern), using long-term memory. This is how neural structures are created. The author also admits that a similar method can be successfully applied to the training of artificial intelligence neural networks. However, this assumption requires further research and verification. The educational method and program proposed by the author meet the modern requirements for education, which involves mastering various areas of knowledge, starting from an early age. This approach makes it possible to involve the child's cognitive potential as much as possible and direct it to the preservation and development of individual talents. According to the methodology, at the early stages of learning students understand the connection between school subjects (so-called "sciences" and "humanities") and in real life, apply the knowledge gained in practice. This approach allows students to realize their natural creative abilities and talents, which makes it easier to navigate professional choices and find their place in life.

Keywords: science education, maths education, AI, neuroplasticity, innovative education problem, creativity development, modern education problem

Procedia PDF Downloads 40

Authors: Varun Agarwal

Abstract:

Introduction: With the advent of personalized medicine, histopathological review of whole slide images (WSIs) for cancer diagnosis presents an exceedingly time-consuming, complex task. Specifically, detecting metastatic regions in WSIs of sentinel lymph node biopsies necessitates a full-scanned, holistic evaluation of the image. Thus, digital pathology, low-level image manipulation algorithms, and machine learning provide significant advancements in improving the efficiency and accuracy of WSI analysis. Using Camelyon16 data, this paper proposes a deep learning pipeline to automate and ameliorate breast cancer metastasis localization and WSI classification. Methodology: The model broadly follows five stages -region of interest detection, WSI partitioning into image tiles, convolutional neural network (CNN) image-segment classifications, probabilistic mapping of tumor localizations, and further processing for whole WSI classification. Transfer learning is applied to the task, with the implementation of Inception-ResNetV2 - an effective CNN classifier that uses residual connections to enhance feature representation, adding convolved outputs in the inception unit to the proceeding input data. Moreover, in order to augment the performance of the transfer learning CNN, a stack of convolutional denoising autoencoders (CDAE) is applied to produce embeddings that enrich image representation. Through a saliency-detection algorithm, visual training segments are generated, which are then processed through a denoising autoencoder -primarily consisting of convolutional, leaky rectified linear unit, and batch normalization layers- and subsequently a contrast-normalization function. A spatial pyramid pooling algorithm extracts the key features from the processed image, creating a viable feature map for the CNN that minimizes spatial resolution and noise. Results and Conclusion: The simplified and effective architecture of the fine-tuned transfer learning Inception-ResNetV2 network enhanced with the CDAE stack yields state of the art performance in WSI classification and tumor localization, achieving AUC scores of 0.947 and 0.753, respectively. The convolutional feature retention and compilation with the residual connections to inception units synergized with the input denoising algorithm enable the pipeline to serve as an effective, efficient tool in the histopathological review of WSIs.

Keywords: breast cancer, convolutional neural networks, metastasis mapping, whole slide images

Procedia PDF Downloads 113

Authors: Maryam Zarabian, Hector Guzman, Pedro Pereira-Almao, Abraham Fapojuwo

Abstract:

Dry reforming of methane (DRM) has sparked significant industrial and scientific interest not only as a viable alternative for addressing the environmental concerns of two main contributors of the greenhouse effect, i.e., carbon dioxide (CO₂) and methane (CH₄), but also produces syngas, i.e., a mixture of hydrogen (H₂) and carbon monoxide (CO) utilized by a wide range of downstream processes as a feedstock for other chemical productions. In this study, we develop an AI-enable syngas production model to tackle the problem of achieving an equivalent H₂/CO ratio [1:1] with respect to the most efficient conversion. Firstly, the unsupervised density-based spatial clustering of applications with noise (DBSAN) algorithm removes outlier data points from the original experimental dataset. Then, random forest (RF) and deep neural network (DNN) models employ the error-free dataset to predict the DRM results. DNN models inherently would not be able to obtain accurate predictions without a huge dataset. To cope with this limitation, we employ reusing pre-trained layers’ approaches such as transfer learning and greedy layer-wise pretraining. Compared to the other deep models (i.e., pure deep model and transferred deep model), the greedy layer-wise pre-trained deep model provides the most accurate prediction as well as similar accuracy to the RF model with R² values 1.00, 0.999, 0.999, 0.999, 0.999, and 0.999 for the total outlet flow, H₂/CO ratio, H₂ yield, CO yield, CH₄ conversion, and CO₂ conversion outputs, respectively.

Keywords: artificial intelligence, dry reforming of methane, artificial neural network, deep learning, machine learning, transfer learning, greedy layer-wise pretraining

Procedia PDF Downloads 68

Authors: Pius Marthin, Nihal Ata Tutkun

Abstract:

Survival analysis has become one of the paramount procedures in the modeling of time-to-event data. When we encounter complex survival problems, the traditional approach remains limited in accounting for the complex correlational structure between the covariates and the outcome due to the strong assumptions that limit the inference and prediction ability of the resulting models. Several studies exist on the deep learning approach to survival modeling; moreover, the application for the case of complex survival problems still needs to be improved. In addition, the existing models need to address the data structure's complexity fully and are subject to noise and redundant information. In this study, we design a deep learning technique (CmpXRnnSurv_AE) that obliterates the limitations imposed by traditional approaches and addresses the above issues to jointly predict the risk-specific probabilities and survival function for recurrent events with competing risks. We introduce the component termed Risks Information Weights (RIW) as an attention mechanism to compute the weighted cumulative incidence function (WCIF) and an external auto-encoder (ExternalAE) as a feature selector to extract complex characteristics among the set of covariates responsible for the cause-specific events. We train our model using synthetic and real data sets and employ the appropriate metrics for complex survival models for evaluation. As benchmarks, we selected both traditional and machine learning models and our model demonstrates better performance across all datasets.

Keywords: cumulative incidence function (CIF), risk information weight (RIW), autoencoders (AE), survival analysis, recurrent events with competing risks, recurrent neural networks (RNN), long short-term memory (LSTM), self-attention, multilayers perceptrons (MLPs)

Procedia PDF Downloads 69

Authors: F. Ghazalnaz Sharifonnasabi, Iman Makhdoom

Abstract:

Breast cancer is a serious health concern that affects many people around the world. According to a study published in the Breast journal, the global burden of breast cancer is expected to increase significantly over the next few decades. The number of deaths from breast cancer has been increasing over the years, but the age-standardized mortality rate has decreased in some countries. It’s important to be aware of the risk factors for breast cancer and to get regular check- ups to catch it early if it does occur. Machin learning techniques have been used to aid in the early detection and diagnosis of breast cancer. These techniques, that have been shown to be effective in predicting and diagnosing the disease, have become a research hotspot. In this study, we consider two deep learning approaches including: Multi-Layer Perceptron (MLP), and Convolutional Neural Network (CNN). We also considered the five-machine learning algorithm titled: Decision Tree (C4.5), Naïve Bayesian (NB), Support Vector Machine (SVM), K-Nearest Neighbors (KNN) Algorithm and XGBoost (eXtreme Gradient Boosting) on the Breast Cancer Wisconsin Diagnostic dataset. We have carried out the process of evaluating and comparing classifiers involving selecting appropriate metrics to evaluate classifier performance and selecting an appropriate tool to quantify this performance. The main purpose of the study is predicting and diagnosis breast cancer, applying the mentioned algorithms and also discovering of the most effective with respect to confusion matrix, accuracy and precision. It is realized that CNN outperformed all other classifiers and achieved the highest accuracy (0.982456). The work is implemented in the Anaconda environment based on Python programing language.

Keywords: breast cancer, multi-layer perceptron, Naïve Bayesian, SVM, decision tree, convolutional neural network, XGBoost, KNN

Procedia PDF Downloads 55

Authors: Jinan Fiaidhi, Sabah Mohammed, Petros Zezos

Abstract:

Capsule endoscopy (CE) is an established noninvasive diagnostic modality in investigating small bowel disease. CE has a pivotal role in assessing patients with suspected bleeding or identifying evidence of active Crohn's disease in the small bowel. However, CE produces lengthy videos with at least eighty thousand frames, with a frequency rate of 2 frames per second. Gastroenterologists cannot dedicate 8 to 15 hours to reading the CE video frames to arrive at a diagnosis. This is why the issue of analyzing CE videos based on modern artificial intelligence techniques becomes a necessity. However, machine learning, including deep learning, has failed to report robust results because of the lack of large samples to train its neural nets. In this paper, we are describing a thick data approach that learns from a few anchor images. We are using sound datasets like KVASIR and CrohnIPI to filter candidate frames that include interesting anomalies in any CE video. We are identifying candidate frames based on feature extraction to provide representative measures of the anomaly, like the size of the anomaly and the color contrast compared to the image background, and later feed these features to a decision tree that can classify the candidate frames as having a condition like the Crohn's Disease. Our thick data approach reported accuracy of detecting Crohn's Disease based on the availability of ulcer areas at the candidate frames for KVASIR was 89.9% and for the CrohnIPI was 83.3%. We are continuing our research to fine-tune our approach by adding more thick data methods for enhancing diagnosis accuracy.

Keywords: thick data analytics, capsule endoscopy, Crohn’s disease, siamese neural network, decision tree

Procedia PDF Downloads 129

Authors: Abeer A. Aljohani

Abstract:

COVID-19 virus spread has been one of the extreme pandemics across the globe. It is also referred to as coronavirus, which is a contagious disease that continuously mutates into numerous variants. Currently, the B.1.1.529 variant labeled as omicron is detected in South Africa. The huge spread of COVID-19 disease has affected several lives and has surged exceptional pressure on the healthcare systems worldwide. Also, everyday life and the global economy have been at stake. This research aims to predict COVID-19 disease in its initial stage to reduce the death count. Machine learning (ML) is nowadays used in almost every area. Numerous COVID-19 cases have produced a huge burden on the hospitals as well as health workers. To reduce this burden, this paper predicts COVID-19 disease is based on the symptoms and medical history of the patient. This research presents a unique architecture for COVID-19 detection using ML techniques integrated with feature dimensionality reduction. This paper uses a standard UCI dataset for predicting COVID-19 disease. This dataset comprises symptoms of 5434 patients. This paper also compares several supervised ML techniques to the presented architecture. The architecture has also utilized 10-fold cross validation process for generalization and the principal component analysis (PCA) technique for feature reduction. Standard parameters are used to evaluate the proposed architecture including F1-Score, precision, accuracy, recall, receiver operating characteristic (ROC), and area under curve (AUC). The results depict that decision tree, random forest, and neural networks outperform all other state-of-the-art ML techniques. This achieved result can help effectively in identifying COVID-19 infection cases.

Keywords: supervised machine learning, COVID-19 prediction, healthcare analytics, random forest, neural network

Procedia PDF Downloads 71

Authors: Syed Asif Hassan, Syed Atif Hassan

Abstract:

Multidrug-resistant Tuberculosis (MDR-TB) is an infection caused by the resistant strains of Mycobacterium tuberculosis that do not respond either to isoniazid or rifampicin, which are the most important anti-TB drugs. The increase in the occurrence of a drug-resistance strain of MTB calls for an intensive search of novel target-based therapeutics. In this context LprG (Rv1411c) a lipoprotein from MTB plays a pivotal role in the immune evasion of Mtb leading to survival and propagation of the bacterium within the host cell. Therefore, a machine learning method will be developed for generating a computational model that could predict for a potential anti LprG activity of the novel antituberculosis compound. The present study will utilize dataset from PubChem database maintained by National Center for Biotechnology Information (NCBI). The dataset involves compounds screened against MTB were categorized as active and inactive based upon PubChem activity score. PowerMV, a molecular descriptor generator, and visualization tool will be used to generate the 2D molecular descriptors for the actives and inactive compounds present in the dataset. The 2D molecular descriptors generated from PowerMV will be used as features. We feed these features into three different classifiers, namely, random forest, a deep neural network, and a recurring neural network, to build separate predictive models and choosing the best performing model based on the accuracy of predicting novel antituberculosis compound with an anti LprG activity. Additionally, the efficacy of predicted active compounds will be screened using SMARTS filter to choose molecule with drug-like features.

Keywords: antituberculosis drug, classifier, machine learning, molecular descriptors, prediction

Procedia PDF Downloads 371

Authors: Goutam Kumar Ghorai, Sandip Sadhukhan, Arpita Sarkar, Debprasad Sinha, G. Sarkar, Ashis K. Dhara

Abstract:

Diabetes mellitus is one of the most common chronic diseases in all countries and continues to increase in numbers significantly. Diabetic retinopathy (DR) is damage to the retina that occurs with long-term diabetes. DR is a major cause of blindness in the Indian population. Therefore, its early diagnosis is of utmost importance towards preventing progression towards imminent irreversible loss of vision, particularly in the huge population across rural India. The barriers to eye examination of all diabetic patients are socioeconomic factors, lack of referrals, poor access to the healthcare system, lack of knowledge, insufficient number of ophthalmologists, and lack of networking between physicians, diabetologists and ophthalmologists. A few diabetic patients often visit a healthcare facility for their general checkup, but their eye condition remains largely undetected until the patient is symptomatic. This work aims to focus on the design and development of a fully automated intelligent decision system for screening retinal fundus images towards detection of the pathophysiology caused by microaneurysm in the early stage of the diseases. Automated detection of microaneurysm is a challenging problem due to the variation in color and the variation introduced by the field of view, inhomogeneous illumination, and pathological abnormalities. We have developed aconvolutional neural network for efficient detection of microaneurysm. A loss function is also developed to handle severe class imbalance due to very small size of microaneurysms compared to background. The network is able to locate the salient region containing microaneurysms in case of noisy images captured by non-mydriatic cameras. The ground truth of microaneurysms is created by expert ophthalmologists for MESSIDOR database as well as private database, collected from Indian patients. The network is trained from scratch using the fundus images of MESSIDOR database. The proposed method is evaluated on DIARETDB1 and the private database. The method is successful in detection of microaneurysms for dilated and non-dilated types of fundus images acquired from different medical centres. The proposed algorithm could be used for development of AI based affordable and accessible system, to provide service at grass root-level primary healthcare units spread across the country to cater to the need of the rural people unaware of the severe impact of DR.

Keywords: retinal fundus image, deep convolutional neural network, early detection of microaneurysms, screening of diabetic retinopathy

Procedia PDF Downloads 125

Authors: Farid Khosravikia, Patricia Clayton

Abstract:

The main objective of this work is to evaluate the advantages and disadvantages of various machine learning techniques in two key steps of seismic hazard and risk assessment of different types of structures. The first step is the development of ground-motion models, which are used for forecasting ground-motion intensity measures (IM) given source characteristics, source-to-site distance, and local site condition for future events. IMs such as peak ground acceleration and velocity (PGA and PGV, respectively) as well as 5% damped elastic pseudospectral accelerations at different periods (PSA), are indicators of the strength of shaking at the ground surface. Typically, linear regression-based models, with pre-defined equations and coefficients, are used in ground motion prediction. However, due to the restrictions of the linear regression methods, such models may not capture more complex nonlinear behaviors that exist in the data. Thus, this study comparatively investigates potential benefits from employing other machine learning techniques as statistical method in ground motion prediction such as Artificial Neural Network, Random Forest, and Support Vector Machine. The results indicate the algorithms satisfy some physically sound characteristics such as magnitude scaling distance dependency without requiring pre-defined equations or coefficients. Moreover, it is shown that, when sufficient data is available, all the alternative algorithms tend to provide more accurate estimates compared to the conventional linear regression-based method, and particularly, Random Forest outperforms the other algorithms. However, the conventional method is a better tool when limited data is available. Second, it is investigated how machine learning techniques could be beneficial for developing probabilistic seismic demand models (PSDMs), which provide the relationship between the structural demand responses (e.g., component deformations, accelerations, internal forces, etc.) and the ground motion IMs. In the risk framework, such models are used to develop fragility curves estimating exceeding probability of damage for pre-defined limit states, and therefore, control the reliability of the predictions in the risk assessment. In this study, machine learning algorithms like artificial neural network, random forest, and support vector machine are adopted and trained on the demand parameters to derive PSDMs for them. It is observed that such models can provide more accurate estimates of prediction in relatively shorter about of time compared to conventional methods. Moreover, they can be used for sensitivity analysis of fragility curves with respect to many modeling parameters without necessarily requiring more intense numerical response-history analysis.

Keywords: artificial neural network, machine learning, random forest, seismic risk analysis, seismic hazard analysis, support vector machine

Procedia PDF Downloads 86

Authors: Oumaima Khlifati, Khadija Baba

Abstract:

Pavement distress is the main factor responsible for the deterioration of road structure durability, damage vehicles, and driver comfort. Transportation agencies spend a high proportion of their funds on pavement monitoring and maintenance. The auscultation of pavement distress was based on the manual survey, which was extremely time consuming, labor intensive, and required domain expertise. Therefore, the automatic distress detection is needed to reduce the cost of manual inspection and avoid more serious damage by implementing the appropriate remediation actions at the right time. Inspired by recent deep learning applications, this paper proposes an algorithm for automatic road distress detection and classification using on the Deep Convolutional Neural Network (DCNN). In this study, the types of pavement distress are classified as transverse or longitudinal cracking, alligator, pothole, and intact pavement. The dataset used in this work is composed of public asphalt pavement images. In order to learn the structure of the different type of distress, the DCNN models are trained and tested as a multi-label classification task. In addition, to get the highest accuracy for our model, we adjust the structural optimization hyper parameters such as the number of convolutions and max pooling, filers, size of filters, loss functions, activation functions, and optimizer and fine-tuning hyper parameters that conclude batch size and learning rate. The optimization of the model is executed by checking all feasible combinations and selecting the best performing one. The model, after being optimized, performance metrics is calculated, which describe the training and validation accuracies, precision, recall, and F1 score.

Keywords: distress pavement, hyperparameters, automatic classification, deep learning

Procedia PDF Downloads 66

Authors: Redvan Ghasemlounia, Elham Ansari, Hikmet Kerem Cigizoglu

Abstract:

Hydrological modeling in arid and semi-arid regions is very important. Iran has many regions with these climate conditions such as Chaharmahal and Bakhtiari province that needs lots of attention with an appropriate management. Forecasting of hydrological parameters and estimation of hydrological events of catchments, provide important information that used for design, management and operation of water resources such as river systems, and dams, widely. Discharge in rivers is one of these parameters. This study presents the application and comparison of some estimation methods such as Feed-Forward Back Propagation Neural Network (FFBPNN), Multi Linear Regression (MLR), Gene Expression Programming (GEP) and Bayesian Network (BN) to predict the daily flow discharge of the Soolegan River, located at Chaharmahal and Bakhtiari province, in Iran. In this study, Soolegan, station was considered. This Station is located in Soolegan River at 51° 14՜ Latitude 31° 38՜ longitude at North Karoon basin. The Soolegan station is 2086 meters higher than sea level. The data used in this study are daily discharge and daily precipitation of Soolegan station. Feed Forward Back Propagation Neural Network(FFBPNN), Multi Linear Regression (MLR), Gene Expression Programming (GEP) and Bayesian Network (BN) models were developed using the same input parameters for Soolegan's daily discharge estimation. The results of estimation models were compared with observed discharge values to evaluate performance of the developed models. Results of all methods were compared and shown in tables and charts.

Keywords: ANN, multi linear regression, Bayesian network, forecasting, discharge, gene expression programming

Procedia PDF Downloads 548

Authors: Jaiden Xuan Schraut, Leon Liu, Yiqiao Yin

Abstract:

Computer vision in medical diagnosis has achieved a high level of success in diagnosing diseases with high accuracy. However, conventional classifiers that produce an image to-label result provides insufficient information for medical professionals to judge and raise concerns over the trust and reliability of a model with results that cannot be explained. In order to gain local insight into cancerous regions, separate tasks such as imaging segmentation need to be implemented to aid the doctors in treating patients, which doubles the training time and costs which renders the diagnosis system inefficient and difficult to be accepted by the public. To tackle this issue and drive AI-first medical solutions further, this paper proposes a multi-output network that follows a U-Net architecture for image segmentation output and features an additional convolutional neural networks (CNN) module for auxiliary classification output. Class activation maps are a method of providing insight into a convolutional neural network’s feature maps that leads to its classification but in the case of lung diseases, the region of interest is enhanced by U-net-assisted Class Activation Map (CAM) visualization. Therefore, our proposed model combines image segmentation models and classifiers to crop out only the lung region of a chest X-ray’s class activation map to provide a visualization that improves the explainability and is able to generate classification results simultaneously which builds trust for AI-led diagnosis systems. The proposed U-Net model achieves 97.61% accuracy and a dice coefficient of 0.97 on testing data from the COVID-QU-Ex Dataset which includes both diseased and healthy lungs.

Keywords: multi-output network model, U-net, class activation map, image classification, medical imaging analysis

Procedia PDF Downloads 176

Authors: V. Veeraprathap, G. S. Harish, G. Narendra Kumar

Abstract:

Uncontrolled growth of abnormal cells in the lung in the form of tumor can be either benign (non-cancerous) or malignant (cancerous). Patients with Lung Cancer (LC) have an average of five years life span expectancy provided diagnosis, detection and prediction, which reduces many treatment options to risk of invasive surgery increasing survival rate. Computed Tomography (CT), Positron Emission Tomography (PET), and Magnetic Resonance Imaging (MRI) for earlier detection of cancer are common. Gaussian filter along with median filter used for smoothing and noise removal, Histogram Equalization (HE) for image enhancement gives the best results without inviting further opinions. Lung cavities are extracted and the background portion other than two lung cavities is completely removed with right and left lungs segmented separately. Region properties measurements area, perimeter, diameter, centroid and eccentricity measured for the tumor segmented image, while texture is characterized by Gray-Level Co-occurrence Matrix (GLCM) functions, feature extraction provides Region of Interest (ROI) given as input to classifier. Two levels of classifications, K-Nearest Neighbor (KNN) is used for determining patient condition as normal or abnormal, while Artificial Neural Networks (ANN) is used for identifying the cancer stage is employed. Discrete Wavelet Transform (DWT) algorithm is used for the main feature extraction leading to best efficiency. The developed technology finds encouraging results for real time information and on line detection for future research.

Keywords: artificial neural networks, ANN, discrete wavelet transform, DWT, gray-level co-occurrence matrix, GLCM, k-nearest neighbor, KNN, region of interest, ROI

Procedia PDF Downloads 130

Authors: Sangita Pokhrel, Nalinda Somasiri, Rebecca Jeyavadhanam, Swathi Ganesan

Abstract:

Tourism is a booming industry with huge future potential for global wealth and employment. There are countless data generated over social media sites every day, creating numerous opportunities to bring more insights to decision-makers. The integration of Big Data Technology into the tourism industry will allow companies to conclude where their customers have been and what they like. This information can then be used by businesses, such as those in charge of managing visitor centers or hotels, etc., and the tourist can get a clear idea of places before visiting. The technical perspective of natural language is processed by analysing the sentiment features of online reviews from tourists, and we then supply an enhanced long short-term memory (LSTM) framework for sentiment feature extraction of travel reviews. We have constructed a web review database using a crawler and web scraping technique for experimental validation to evaluate the effectiveness of our methodology. The text form of sentences was first classified through Vader and Roberta model to get the polarity of the reviews. In this paper, we have conducted study methods for feature extraction, such as Count Vectorization and TFIDF Vectorization, and implemented Convolutional Neural Network (CNN) classifier algorithm for the sentiment analysis to decide the tourist’s attitude towards the destinations is positive, negative, or simply neutral based on the review text that they posted online. The results demonstrated that from the CNN algorithm, after pre-processing and cleaning the dataset, we received an accuracy of 96.12% for the positive and negative sentiment analysis.

Keywords: counter vectorization, convolutional neural network, crawler, data technology, long short-term memory, web scraping, sentiment analysis

Procedia PDF Downloads 72

Authors: Justin Durham, Chanda Rooney, Robert Mather, Mickie Vanhoy

Abstract:

The psychological construct of empathy is to understand a person’s cognitive perspective and experience the other person’s emotional state. Deciphering emotional states is conducive for interpreting vicarious pain. Observing others' physical pain activates neural networks related to the actual experience of pain itself. The study addresses empathy as a nonlinear dynamic process of simulation for individuals to understand the mental states of others and experience vicarious pain, exhibiting self-organized criticality. Such criticality follows from a combination of neural networks with an excitatory feedback loop generating bistability to resonate permutated empathy. Cortical networks exhibit diverse patterns of activity, including oscillations, synchrony and waves, however, the temporal dynamics of neurophysiological activities underlying empathic processes remain poorly understood. Mu rhythms are EEG oscillations with dominant frequencies of 8-13 Hz becoming synchronized when the body is relaxed with eyes open and when the sensorimotor system is in idle, thus, mu rhythm synchrony is expected to be highest in baseline conditions. When the sensorimotor system is activated either by performing or simulating action, mu rhythms become suppressed or desynchronize, thus, should be suppressed while observing video clips of painful injuries if previous research on mirror system activation holds. Twelve undergraduates contributed EEG data and survey responses to empathy and psychopathy scales in addition to watching consecutive video clips of sports injuries. Participants watched a blank, black image on a computer monitor before and after observing a video of consecutive sports injuries incidents. Each video condition lasted five-minutes long. A BIOPAC MP150 recorded EEG signals from sensorimotor and thalamocortical regions related to a complex neural network called the ‘pain matrix’. Physical and social pain are activated in this network to resonate vicarious pain responses to processing empathy. Five EEG single electrode locations were applied to regions measuring sensorimotor electrical activity in microvolts (μV) to monitor mu rhythms. EEG signals were sampled at a rate of 200 Hz. Mu rhythm desynchronization was measured via 8-13 Hz at electrode sites (F3 & F4). Data for each participant’s mu rhythms were analyzed via Fast Fourier Transformation (FFT) and multifractal time series analysis.

Keywords: desynchronization, dynamical systems theory, electroencephalography (EEG), empathy, multifractal time series analysis, mu waveform, neurophysiology, pain simulation, social cognition

Procedia PDF Downloads 270

Authors: Sergio A. Navarro Tuch, Martin Rogelio Bustamante Bello, Leopoldo Julian Lechuga Lopez

Abstract:

In recent years, Mexico’s population has seen a rise of different physiological and mental negative states. Two main consequences of this problematic are deficient work performance and high levels of stress generating and important impact on a person’s physical, mental and emotional health. Several approaches, such as the use of audiovisual stimulus to induce emotions and modify a person’s emotional state, can be applied in an effort to decreases these negative effects. With the use of different non-invasive physiological sensors such as EEG, luminosity and face recognition we gather information of the subject’s current emotional state. In a controlled environment, a subject is shown a series of selected images from the International Affective Picture System (IAPS) in order to induce a specific set of emotions and obtain information from the sensors. The raw data obtained is statistically analyzed in order to filter only the specific groups of information that relate to a subject’s emotions and current values of the physical variables in the controlled environment such as, luminosity, RGB light color, temperature, oxygen level and noise. Finally, a neural network based control algorithm is given the data obtained in order to feedback the system and automate the modification of the environment variables and audiovisual content shown in an effort that these changes can positively alter the subject’s emotional state. During the research, it was found that the light color was directly related to the type of impact generated by the audiovisual content on the subject’s emotional state. Red illumination increased the impact of violent images and green illumination along with relaxing images decreased the subject’s levels of anxiety. Specific differences between men and women were found as to which type of images generated a greater impact in either gender. The population sample was mainly constituted by college students whose data analysis showed a decreased sensibility to violence towards humans. Despite the early stage of the control algorithm, the results obtained from the population sample give us a better insight into the possibilities of emotional domotics and the applications that can be created towards the improvement of performance in people’s lives. The objective of this research is to create a positive impact with the application of technology to everyday activities; nonetheless, an ethical problem arises since this can also be applied to control a person’s emotions and shift their decision making.

Keywords: data analysis, emotional domotics, performance improvement, neural network

Procedia PDF Downloads 121

Authors: Zouari Ezzeddine, Ghraieb Ikram

Abstract:

From the results of the academic literature which evokes the limitations of previous studies, this article shows the reasons for multidimensionality Prediction of financial bubbles. A new framework for modeling study predicting financial bubbles by linking a set of variable presented on several dimensions dictating its multidimensional character. It takes into account the preferences of financial actors. A multicriteria anticipation of the appearance of bubbles in international financial markets helps to fight against a possible crisis.

Keywords: classical measures, predictions, financial bubbles, multidimensional, artificial neural networks

Procedia PDF Downloads 553

Authors: Hozaifa Zaki, Ghada Soliman

Abstract:

In this paper, we introduced an approach for Automated Dotted Arabic Expiration Date Extraction using Optimized Convolutional Autoencoder (ARABEX) with bidirectional LSTM. This approach is used for translating the Arabic dot-matrix expiration dates into their corresponding filled-in dates. A custom lightweight Convolutional Recurrent Neural Network (CRNN) model is then employed to extract the expiration dates. Due to the lack of available dataset images for the Arabic dot-matrix expiration date, we generated synthetic images by creating an Arabic dot-matrix True Type Font (TTF) matrix to address this limitation. Our model was trained on a realistic synthetic dataset of 3287 images, covering the period from 2019 to 2027, represented in the format of yyyy/mm/dd. We then trained our custom CRNN model using the generated synthetic images to assess the performance of our model (ARABEX) by extracting expiration dates from the translated images. Our proposed approach achieved an accuracy of 99.4% on the test dataset of 658 images, while also achieving a Structural Similarity Index (SSIM) of 0.46 for image translation on our dataset. The ARABEX approach demonstrates its ability to be applied to various downstream learning tasks, including image translation and reconstruction. Moreover, this pipeline (ARABEX+CRNN) can be seamlessly integrated into automated sorting systems to extract expiry dates and sort products accordingly during the manufacturing stage. By eliminating the need for manual entry of expiration dates, which can be time-consuming and inefficient for merchants, our approach offers significant results in terms of efficiency and accuracy for Arabic dot-matrix expiration date recognition.

Keywords: computer vision, deep learning, image processing, character recognition

Procedia PDF Downloads 59

Authors: Kaleem Kashif, Tayyaba Anam, Yizhi Wu

Abstract:

This paper introduces a methodology for advancing Italian speech vowels landmark detection within the distinctive feature-based speech recognition domain. Leveraging the legacy tool 'xkl' by integrating combined convolutional neural networks (CNNs) and recurrent neural networks (RNNs), the study presents a comprehensive enhancement to the 'xkl' legacy software. This integration incorporates re-assigned spectrogram methodologies, enabling meticulous acoustic analysis. Simultaneously, our proposed model, integrating combined CNNs and RNNs, demonstrates unprecedented precision and robustness in landmark detection. The augmentation of re-assigned spectrogram fusion within the 'xkl' software signifies a meticulous advancement, particularly enhancing precision related to vowel formant estimation. This augmentation catalyzes unparalleled accuracy in landmark detection, resulting in a substantial performance leap compared to conventional methods. The proposed model emerges as a state-of-the-art solution in the distinctive feature-based speech recognition systems domain. In the realm of deep learning, a synergistic integration of combined CNNs and RNNs is introduced, endowed with specialized temporal embeddings, harnessing self-attention mechanisms, and positional embeddings. The proposed model allows it to excel in capturing intricate dependencies within Italian speech vowels, rendering it highly adaptable and sophisticated in the distinctive feature domain. Furthermore, our advanced temporal modeling approach employs Bayesian temporal encoding, refining the measurement of inter-landmark intervals. Comparative analysis against state-of-the-art models reveals a substantial improvement in accuracy, highlighting the robustness and efficacy of the proposed methodology. Upon rigorous testing on a database (LaMIT) speech recorded in a silent room by four Italian native speakers, the landmark detector demonstrates exceptional performance, achieving a 95% true detection rate and a 10% false detection rate. A majority of missed landmarks were observed in proximity to reduced vowels. These promising results underscore the robust identifiability of landmarks within the speech waveform, establishing the feasibility of employing a landmark detector as a front end in a speech recognition system. The synergistic integration of re-assigned spectrogram fusion, CNNs, RNNs, and Bayesian temporal encoding not only signifies a significant advancement in Italian speech vowels landmark detection but also positions the proposed model as a leader in the field. The model offers distinct advantages, including unparalleled accuracy, adaptability, and sophistication, marking a milestone in the intersection of deep learning and distinctive feature-based speech recognition. This work contributes to the broader scientific community by presenting a methodologically rigorous framework for enhancing landmark detection accuracy in Italian speech vowels. The integration of cutting-edge techniques establishes a foundation for future advancements in speech signal processing, emphasizing the potential of the proposed model in practical applications across various domains requiring robust speech recognition systems.

Keywords: landmark detection, acoustic analysis, convolutional neural network, recurrent neural network

Procedia PDF Downloads 38

Authors: Abdesselem Dakhli, Wajdi Bellil, Chokri Ben Amar

Abstract:

DNA Barcode, a short mitochondrial DNA fragment, made up of three subunits; a phosphate group, sugar and nucleic bases (A, T, C, and G). They provide good sources of information needed to classify living species. Such intuition has been confirmed by many experimental results. Species classification with DNA Barcode sequences has been studied by several researchers. The classification problem assigns unknown species to known ones by analyzing their Barcode. This task has to be supported with reliable methods and algorithms. To analyze species regions or entire genomes, it becomes necessary to use similarity sequence methods. A large set of sequences can be simultaneously compared using Multiple Sequence Alignment which is known to be NP-complete. To make this type of analysis feasible, heuristics, like progressive alignment, have been developed. Another tool for similarity search against a database of sequences is BLAST, which outputs shorter regions of high similarity between a query sequence and matched sequences in the database. However, all these methods are still computationally very expensive and require significant computational infrastructure. Our goal is to build predictive models that are highly accurate and interpretable. This method permits to avoid the complex problem of form and structure in different classes of organisms. On empirical data and their classification performances are compared with other methods. Our system consists of three phases. The first is called transformation, which is composed of three steps; Electron-Ion Interaction Pseudopotential (EIIP) for the codification of DNA Barcodes, Fourier Transform and Power Spectrum Signal Processing. The second is called approximation, which is empowered by the use of Multi Llibrary Wavelet Neural Networks (MLWNN).The third is called the classification of DNA Barcodes, which is realized by applying the algorithm of hierarchical classification.

Keywords: DNA barcode, electron-ion interaction pseudopotential, Multi Library Wavelet Neural Networks (MLWNN)

Procedia PDF Downloads 299

Authors: Przemysław Adamczyk, Mirosław Wyczesany, Aleksandra Domagalik, Artur Daren, Kamil Cepuch, Piotr Błądziński, Tadeusz Marek, Andrzej Cechnicki

Abstract:

The present study aimed at evaluation of neural correlates of humor comprehension impairments observed in schizophrenia. To investigate the nature of this deficit in schizophrenia and to localize cortical areas involved in humor processing we used functional magnetic resonance imaging (fMRI). The study included chronic schizophrenia outpatients (SCH; n=20), and sex, age and education level matched healthy controls (n=20). The task consisted of 60 stories (setup) of which 20 had funny, 20 nonsensical and 20 neutral (not funny) punchlines. After the punchlines were presented, the participants were asked to indicate whether the story was comprehensible (yes/no) and how funny it was (1-9 Likert-type scale). fMRI was performed on a 3T scanner (Magnetom Skyra, Siemens) using 32-channel head coil. Three contrasts in accordance with the three stages of humor processing were analyzed in both groups: abstract vs neutral stories - incongruity detection; funny vs abstract - incongruity resolution; funny vs neutral - elaboration. Additionally, parametric modulation analysis was performed using both subjective ratings separately in order to further differentiate the areas involved in incongruity resolution processing. Statistical analysis for behavioral data used U Mann-Whitney test and Bonferroni’s correction, fMRI data analysis utilized whole-brain voxel-wise t-tests with 10-voxel extent threshold and with Family Wise Error (FWE) correction at alpha = 0.05, or uncorrected at alpha = 0.001. Between group comparisons revealed that the SCH subjects had attenuated activation in: the right superior temporal gyrus in case of irresolvable incongruity processing of nonsensical puns (nonsensical > neutral); the left medial frontal gyrus in case of incongruity resolution processing of funny puns (funny > nonsensical) and the interhemispheric ACC in case of elaboration of funny puns (funny > neutral). Additionally, the SCH group revealed weaker activation during funniness ratings in the left ventro-medial prefrontal cortex, the medial frontal gyrus, the angular and the supramarginal gyrus, and the right temporal pole. In comprehension ratings the SCH group showed suppressed activity in the left superior and medial frontal gyri. Interestingly, these differences were accompanied by protraction of time in both types of rating responses in the SCH group, a lower level of comprehension for funny punchlines and a higher funniness for absurd punchlines. Presented results indicate that, in comparison to healthy controls, schizophrenia is characterized by difficulties in humor processing revealed by longer reaction times, impairments of understanding jokes and finding nonsensical punchlines more funny. This is accompanied by attenuated brain activations, especially in the left fronto-parietal and the right temporal cortices. Disturbances of the humor processing seem to be impaired at the all three stages of the humor comprehension process, from incongruity detection, through its resolution to elaboration. The neural correlates revealed diminished neural activity of the schizophrenia brain, as compared with the control group. The study was supported by the National Science Centre, Poland (grant no 2014/13/B/HS6/03091).

Keywords: communication skills, functional magnetic resonance imaging, humor, schizophrenia

Procedia PDF Downloads 197

Authors: Hien Thi Dieu Truong, Mahmoud Al-Sarayreh, Pullanagari Reddy, Marlon M. Reis, Richard Archer

Abstract:

New Zealand mānuka honey is a honeybee product derived mainly from Leptospermum scoparium nectar. The potent antibacterial activity of mānuka honey derives principally from methylglyoxal (MGO), in addition to the hydrogen peroxide and other lesser activities present in all honey. MGO is formed from dihydroxyacetone (DHA) unique to L. scoparium nectar. Mānuka honey also has an idiosyncratic phenolic profile that is useful as a chemical maker. Authentic mānuka honey is highly valuable, but almost all honey is formed from natural mixtures of nectars harvested by a hive over a time period. Once diluted by other nectars, mānuka honey irrevocably loses value. We aimed to apply hyperspectral imaging to honey frames before bulk extraction to minimise the dilution of genuine mānuka by other honey and ensure authenticity at the source. This technology is non-destructive and suitable for an industrial setting. Chemometrics using linear Partial Least Squares (PLS) and Support Vector Machine (SVM) showed limited efficacy in interpreting chemical footprints due to large non-linear relationships between predictor and predictand in a large sample set, likely due to honey quality variability across geographic regions. Therefore, an advanced modelling approach, one-dimensional convolutional neural networks (1D-CNN), was investigated for analysing hyperspectral data for extraction of biochemical information from honey. The 1D-CNN model showed superior prediction of honey quality (R² = 0.73, RMSE = 2.346, RPD= 2.56) to PLS (R² = 0.66, RMSE = 2.607, RPD= 1.91) and SVM (R² = 0.67, RMSE = 2.559, RPD=1.98). Classification of mono-floral manuka honey from multi-floral and non-manuka honey exceeded 90% accuracy for all models tried. Overall, this study reveals the potential of HSI and deep learning modelling for automating the evaluation of honey quality in frames.

Keywords: mānuka honey, quality, purity, potency, deep learning, 1D-CNN, chemometrics

Procedia PDF Downloads 117

Authors: E. Nakamachi, S. Tanaka, K. Yamamoto, Y. Morita

Abstract:

In this study, we developed a three-dimensional (3D) direct current electric field (DCEF) stimulation bio-reactor for axonal outgrowth enhancement to generate the neural network of the central nervous system (CNS). By using our newly developed 3D DCEF stimulation bio-reactor, we cultured the rat pheochromocytoma cells (PC12) and investigated the effects on the axonal extension enhancement and network generation. Firstly, we designed and fabricated a 3D bio-reactor, which can load DCEF stimulation on PC12 cells embedded in the collagen gel as extracellular environment. The connection between the electrolyte and the medium using salt bridges for DCEF stimulation was introduced to avoid the cell death by the toxicity of metal ion. The distance between the salt bridges was adopted as the design variable to optimize a structure for uniform DCEF stimulation, where the finite element (FE) analyses results were used. Uniform DCEF strength and electric flux vector direction in the PC12 cells embedded in collagen gel were examined through measurements of the fabricated 3D bio-reactor chamber. Measurement results of DCEF strength in the bio-reactor showed a good agreement with FE results. In addition, the perfusion system was attached to maintain pH 7.2 ~ 7.6 of the medium because pH change was caused by DCEF stimulation loading. Secondly, we disseminated PC12 cells in collagen gel and carried out 3D culture. Finally, we measured the morphology of PC12 cell bodies and neurites by the multiphoton excitation fluorescence microscope (MPM). The effectiveness of DCEF stimulation to enhance the axonal outgrowth and the neural network generation was investigated. We confirmed that both an increase of mean axonal length and axogenesis rate of PC12, which have been exposed 5 mV/mm for 6 hours a day for 4 days in the bioreactor. We found following conclusions in our study. 1) Design and fabrication of DCEF stimulation bio-reactor capable of 3D culture nerve cell were completed. A uniform electric field strength of average value of 17 mV/mm within the 1.2% error range was confirmed by using FE analyses, after the structure determination through the optimization process. In addition, we attached a perfusion system capable of suppressing the pH change of the culture solution due to DCEF stimulation loading. 2) Evaluation of DCEF stimulation effects on PC12 cell activity was executed. The 3D culture of PC 12 was carried out adopting the embedding culture method using collagen gel as a scaffold for four days under the condition of 5.0 mV/mm and 10mV/mm. There was a significant effect on the enhancement of axonal extension, as 11.3% increase in an average length, and the increase of axogenesis rate. On the other hand, no effects on the orientation of axon against the DCEF flux direction was observed. Further, the network generation was enhanced to connect longer distance between the target neighbor cells by DCEF stimulation.

Keywords: PC12, DCEF stimulation, 3D bio-reactor, axonal extension, neural network generation

Procedia PDF Downloads 173

Authors: Xiang Zhang, Daohong Yang, Ziyuan Wu, Lei Li, Wanting Zhou

Abstract:

Visual Simultaneous Localization and Mapping (VSLAM) uses cameras to collect information in unknown environments to realize simultaneous localization and environment map construction, which has a wide range of applications in autonomous driving, virtual reality and other related fields. At present, the related research achievements about VSLAM can maintain high accuracy in static environment. But in dynamic environment, due to the presence of moving objects in the scene, the movement of these objects will reduce the stability of VSLAM system, resulting in inaccurate localization and mapping, or even failure. In this paper, a robust VSLAM method was proposed to effectively deal with the problem in dynamic environment. We proposed a dynamic region removal scheme based on semantic segmentation neural networks and geometric constraints. Firstly, semantic extraction neural network is used to extract prior active motion region, prior static region and prior passive motion region in the environment. Then, the light weight frame tracking module initializes the transform pose between the previous frame and the current frame on the prior static region. A motion consistency detection module based on multi-view geometry and scene flow is used to divide the environment into static region and dynamic region. Thus, the dynamic object region was successfully eliminated. Finally, only the static region is used for tracking thread. Our research is based on the ORBSLAM3 system, which is one of the most effective VSLAM systems available. We evaluated our method on the TUM RGB-D benchmark and the results demonstrate that the proposed VSLAM method improves the accuracy of the original ORBSLAM3 by 70%˜98.5% under high dynamic environment.

Keywords: dynamic scene, dynamic visual SLAM, semantic segmentation, scene flow, VSLAM

Procedia PDF Downloads 92

Authors: Michael Moeti, Khuliso Sigama, Thapelo Samuel Matlala

Abstract:

One of the principal challenges that the world is confronted with is insecurity. The crime rate is increasing exponentially, and protecting our physical assets especially in the motorist industry, is becoming impossible when applying our own strength. The need to develop technological solutions that detect and report theft without any human interference is inevitable. This is critical, especially for vehicle owners, to ensure theft detection and speedy identification towards recovery efforts in cases where a vehicle is missing or attempted theft is taking place. The vehicle theft detection system uses Convolutional Neural Network (CNN) to recognize the driver's face captured using an installed mobile phone device. The location identification function uses a Global Positioning System (GPS) to determine the real-time location of the vehicle. Upon identification of the location, Global System for Mobile Communications (GSM) technology is used to report or notify the vehicle owner about the whereabouts of the vehicle. The installed mobile app was implemented by making use of python as it is undoubtedly the best choice in machine learning. It allows easy access to machine learning algorithms through its widely developed library ecosystem. The graphical user interface was developed by making use of JAVA as it is better suited for mobile development. Google's online database (Firebase) was used as a means of storage for the application. The system integration test was performed using a simple percentage analysis. Sixty (60) vehicle owners participated in this study as a sample, and questionnaires were used in order to establish the acceptability of the system developed. The result indicates the efficiency of the proposed system, and consequently, the paper proposes the use of the system can effectively monitor the vehicle at any given place, even if it is driven outside its normal jurisdiction. More so, the system can be used as a database to detect, locate and report missing vehicles to different security agencies.

Keywords: CNN, location identification, tracking, GPS, GSM

Procedia PDF Downloads 139

Authors: Chao-Ton Su, Li-Fei Chen

Abstract:

The parameter design is crucial to improving the uniformity of a product or process. In the product design stage, parameter design aims to determine the optimal settings for the parameters of each element in the system, thereby minimizing the functional deviations of the product. In the process design stage, parameter design aims to determine the operating settings of the manufacturing processes so that non-uniformity in manufacturing processes can be minimized. The parameter design, trying to minimize the influence of noise on the manufacturing system, plays an important role in the high-tech companies. Taiwan has many well-known high-tech companies, which show key roles in the global economy. Quality remains the most important factor that enables these companies to sustain their competitive advantage. In Taiwan however, many high-tech companies face various quality problems. A common challenge is related to root causes and defect patterns. In the R&D stage, root causes are often unknown, and defect patterns are difficult to classify. Additionally, data collection is not easy. Even when high-volume data can be collected, data interpretation is difficult. To overcome these challenges, high-tech companies in Taiwan use more advanced quality improvement tools. In addition to traditional statistical methods and quality tools, the new trend is the application of powerful tools, such as neural network, fuzzy theory, data mining, industrial engineering, operations research, and innovation skills. In this study, several examples of optimizing the parameter settings for the manufacturing process in Taiwan’s tech companies will be presented to illustrate proposed approach’s effectiveness. Finally, a discussion of using traditional experimental design versus the proposed approach for process optimization will be made.

Keywords: quality engineering, parameter design, neural network, genetic algorithm, experimental design

Procedia PDF Downloads 128

Authors: Foteini Zagklavara, Peter Jimack, Nikil Kapur, Ozz Querin, Harvey Thompson

Abstract:

The invention and development of the Polymerase Chain Reaction (PCR) technology have revolutionised molecular biology and molecular diagnostics. There is an urgent need to optimise their performance of those devices while reducing the total construction and operation costs. The present study proposes a CFD-enabled optimisation methodology for continuous flow (CF) PCR devices with serpentine-channel structure, which enables the trade-offs between competing objectives of DNA amplification efficiency and pressure drop to be explored. This is achieved by using a surrogate-enabled optimisation approach accounting for the geometrical features of a CF μPCR device by performing a series of simulations at a relatively small number of Design of Experiments (DoE) points, with the use of COMSOL Multiphysics 5.4. The values of the objectives are extracted from the CFD solutions, and response surfaces created using the polyharmonic splines and neural networks. After creating the respective response surfaces, genetic algorithm, and a multi-level coordinate search optimisation function are used to locate the optimum design parameters. Both optimisation methods produced similar results for both the neural network and the polyharmonic spline response surfaces. The results indicate that there is the possibility of improving the DNA efficiency by ∼2% in one PCR cycle when doubling the width of the microchannel to 400 μm while maintaining the height at the value of the original design (50μm). Moreover, the increase in the width of the serpentine microchannel is combined with a decrease in its total length in order to obtain the same residence times in all the simulations, resulting in a smaller total substrate volume (32.94% decrease). A multi-objective optimisation is also performed with the use of a Pareto Front plot. Such knowledge will enable designers to maximise the amount of DNA amplified or to minimise the time taken throughout thermal cycling in such devices.

Keywords: PCR, optimisation, microfluidics, COMSOL

Procedia PDF Downloads 141

Authors: Mathew Wakefield, Matthew Mitchell, Lisa Wise, Christopher McCarthy

Abstract:

The properties of memory representations in artificial neural networks have cognitive implications. Distributed representations that encode instances as a pattern of activity across layers of nodes afford memory compression and enforce the selection of a single point in instance space. These encoding schemes also appear to distort the representational space, as well as trading off the ability to validate that input information is within the bounds of past experience. In contrast, a localist representation which encodes some meaningful information into individual nodes in a network layer affords less memory compression while retaining the integrity of the representational space. This allows the validity of an input to be determined. The validity (or familiarity) of input along with the capacity of localist representation for multiple instance selections affords a memory sampling approach that dynamically balances the bias-variance trade-off. When the input is familiar, bias may be high by referring only to the most similar instances in memory. When the input is less familiar, variance can be increased by referring to more instances that capture a broader range of features. Using this approach in a localist instance memory network, an experiment demonstrates a relationship between representational conflict, generalization performance, and memorization demand. Relatively small sampling ranges produce the best performance on a classic machine learning dataset of visual objects. Combining memory validity with conflict detection produces a reliable confidence judgement that can separate responses with high and low error rates. Confidence can also be used to signal the need for supervisory input. Using this judgement, the need for supervised learning as well as memory encoding can be substantially reduced with only a trivial detriment to classification performance.

Keywords: artificial neural networks, representation, memory, conflict monitoring, confidence

Procedia PDF Downloads 111

Authors: Christine F. Boos, Fernando M. Azevedo

Abstract:

Electroencephalogram (EEG) is a record of the electrical activity of the brain that has many applications, such as monitoring alertness, coma and brain death; locating damaged areas of the brain after head injury, stroke and tumor; monitoring anesthesia depth; researching physiology and sleep disorders; researching epilepsy and localizing the seizure focus. Epilepsy is a chronic condition, or a group of diseases of high prevalence, still poorly explained by science and whose diagnosis is still predominantly clinical. The EEG recording is considered an important test for epilepsy investigation and its visual analysis is very often applied for clinical confirmation of epilepsy diagnosis. Moreover, this EEG analysis can also be used to help define the types of epileptic syndrome, determine epileptiform zone, assist in the planning of drug treatment and provide additional information about the feasibility of surgical intervention. In the context of diagnosis confirmation the analysis is made using long term EEG recordings with at least 24 hours long and acquired by a minimum of 24 electrodes in which the neurophysiologists perform a thorough visual evaluation of EEG screens in search of specific electrographic patterns called epileptiform discharges. Considering that the EEG screens usually display 10 seconds of the recording, the neurophysiologist has to evaluate 360 screens per hour of EEG or a minimum of 8,640 screens per long term EEG recording. Analyzing thousands of EEG screens in search patterns that have a maximum duration of 200 ms is a very time consuming, complex and exhaustive task. Because of this, over the years several studies have proposed automated methodologies that could facilitate the neurophysiologists’ task of identifying epileptiform discharges and a large number of methodologies used neural networks for the pattern classification. One of the differences between all of these methodologies is the type of input stimuli presented to the networks, i.e., how the EEG signal is introduced in the network. Five types of input stimuli have been commonly found in literature: raw EEG signal, morphological descriptors (i.e. parameters related to the signal’s morphology), Fast Fourier Transform (FFT) spectrum, Short-Time Fourier Transform (STFT) spectrograms and Wavelet Transform features. This study evaluates the application of these five types of input stimuli and compares the classification results of neural networks that were implemented using each of these inputs. The performance of using raw signal varied between 43 and 84% efficiency. The results of FFT spectrum and STFT spectrograms were quite similar with average efficiency being 73 and 77%, respectively. The efficiency of Wavelet Transform features varied between 57 and 81% while the descriptors presented efficiency values between 62 and 93%. After simulations we could observe that the best results were achieved when either morphological descriptors or Wavelet features were used as input stimuli.

Keywords: Artificial neural network, electroencephalogram signal, pattern recognition, signal processing

Procedia PDF Downloads 510