Search results for: convolutional neural net

Authors: Vikrant Gupta, Amrit Goswami

Abstract:

The fixed income market forms the basis of the modern financial market. All other assets in financial markets derive their value from the bond market. Owing to its over-the-counter nature, corporate bonds have relatively less data publicly available and thus is researched upon far less compared to Equities. Bond price prediction is a complex financial time series forecasting problem and is considered very crucial in the domain of finance. The bond prices are highly volatile and full of noise which makes it very difficult for traditional statistical time-series models to capture the complexity in series patterns which leads to inefficient forecasts. To overcome the inefficiencies of statistical models, various machine learning techniques were initially used in the literature for more accurate forecasting of time-series. However, simple machine learning methods such as linear regression, support vectors, random forests fail to provide efficient results when tested on highly complex sequences such as stock prices and bond prices. hence to capture these intricate sequence patterns, various deep learning-based methodologies have been discussed in the literature. In this study, a recurrent neural network-based deep learning model using long short term networks for prediction of corporate bond prices has been discussed. Long Short Term networks (LSTM) have been widely used in the literature for various sequence learning tasks in various domains such as machine translation, speech recognition, etc. In recent years, various studies have discussed the effectiveness of LSTMs in forecasting complex time-series sequences and have shown promising results when compared to other methodologies. LSTMs are a special kind of recurrent neural networks which are capable of learning long term dependencies due to its memory function which traditional neural networks fail to capture. In this study, a simple LSTM, Stacked LSTM and a Masked LSTM based model has been discussed with respect to varying input sequences (three days, seven days and 14 days). In order to facilitate faster learning and to gradually decompose the complexity of bond price sequence, an Empirical Mode Decomposition (EMD) has been used, which has resulted in accuracy improvement of the standalone LSTM model. With a variety of Technical Indicators and EMD decomposed time series, Masked LSTM outperformed the other two counterparts in terms of prediction accuracy. To benchmark the proposed model, the results have been compared with traditional time series models (ARIMA), shallow neural networks and above discussed three different LSTM models. In summary, our results show that the use of LSTM models provide more accurate results and should be explored more within the asset management industry.

Keywords: bond prices, long short-term memory, time series forecasting, empirical mode decomposition

Procedia PDF Downloads 137

Authors: Helen Pushkarskaya, David Tolin, Lital Ruderman, Ariel Kirshenbaum, J. MacLaren Kelly, Christopher Pittenger, Ifat Levy

Abstract:

Obsessive-Compulsive Disorder (OCD) produces profound morbidity. Difficulties with decision making and intolerance of uncertainty are prominent clinical features of OCD. The nature and etiology of these deficits are poorly understood. We used a well-validated choice task, grounded in behavioral economic theory, to investigate differences in valuation and value-based choice during decision making under uncertainty in 20 unmedicated participants with OCD and 20 matched healthy controls. Participants’ choices were used to assess individual decision-making characteristics. Compared to controls, individuals with OCD were less consistent in their choices and less able to identify options that were unambiguously preferable. These differences correlated with symptom severity. OCD participants did not differ from controls in how they valued uncertain options when outcome probabilities were known (risk) but were more likely than controls to avoid uncertain options when these probabilities were imprecisely specified (ambiguity). These results suggest that the underlying neural mechanisms of valuation and value-based choices during decision-making are abnormal in OCD. Individuals with OCD show elevated intolerance of uncertainty, but only when outcome probabilities are themselves uncertain. Future research focused on the neural valuation network, which is implicated in value-based computations, may provide new neurocognitive insights into the pathophysiology of OCD. Deficits in decision-making processes may represent a target for therapeutic intervention.

Keywords: obsessive compulsive disorder, decision-making, uncertainty intolerance, risk aversion, ambiguity aversion, valuation

Procedia PDF Downloads 616

Authors: B. Siva Kumar Reddy, B. Lakshmi

Abstract:

Cognitive Radio is a turning out technology that empowers viable usage of the spectrum. Energy Detector-based Sensing is the most broadly utilized spectrum sensing strategy. Besides, it is a lot of generic as receivers does not like any information on the primary user's signals, channel data, of even the sort of modulation. This paper puts forth the execution of energy detection sensing for AM (Amplitude Modulated) signal at 710 KHz, FM (Frequency Modulated) signal at 103.45 MHz (local station frequency), Wi-Fi signal at 2.4 GHz and WiMAX signals at 6 GHz. The OFDM/OFDMA based WiMAX physical layer with convolutional channel coding is actualized utilizing USRP N210 (Universal Software Radio Peripheral) and GNU Radio based Software Defined Radio (SDR). Test outcomes demonstrated the BER (Bit Error Rate) augmentation with channel noise and BER execution is dissected for different Eb/N0 (the energy per bit to noise power spectral density ratio) values.

Keywords: BER, Cognitive Radio, GNU Radio, OFDM, SDR, WiMAX

Procedia PDF Downloads 503

Authors: Milica Karadzic, Lidija Jevric, Sanja Podunavac-Kuzmanovic, Strahinja Kovacevic, Anamarija Mandic, Katarina Penov Gasi, Marija Sakac, Aleksandar Okljesa, Andrea Nikolic

Abstract:

The present paper deals with chromatographic lipophilicity prediction of newly synthesized steroid derivatives. The prediction was achieved using in silico generated molecular descriptors and quantitative structure-retention relationship (QSRR) methodology with the artificial neural networks (ANN) approach. Chromatographic lipophilicity of the investigated compounds was expressed as retention factor value logk. For QSRR modeling, a feedforward back-propagation ANN with gradient descent learning algorithm was applied. Using the novel sum of ranking differences (SRD) method generated ANN models were ranked. The aim was to distinguish the most consistent QSRR model that can be found, and similarity or dissimilarity between the models that could be noticed. In this study, SRD was performed with average values of retention factor value logk as reference values. An excellent correlation between experimentally observed retention factor value logk and values predicted by the ANN was obtained with a correlation coefficient higher than 0.9890. Statistical results show that the established ANN models can be applied for required purpose. This article is based upon work from COST Action (TD1305), supported by COST (European Cooperation in Science and Technology).

Keywords: artificial neural networks, liquid chromatography, molecular descriptors, steroids, sum of ranking differences

Procedia PDF Downloads 322

Authors: Bahaa Eldin A. Fouda, Khaled N. Yossef, Mohamed Elhosseny, Ahmed Lotfy, Mohamed Salama, Mohamed Sobh

Abstract:

Developmental neurotoxicity (DNT) entails the toxic effects imparted by various chemicals on the brain during the early childhood period. As human brains are vulnerable during this period, various chemicals would have their maximum effects on brains during early childhood. Some toxicants have been confirmed to induce developmental toxic effects on CNS e.g. lead, however; most of the agents cannot be identified with certainty due the defective nature of predictive toxicology models used. A novel alternative method that can overcome most of the limitations of conventional techniques is the use of 3D neurospheres system. This in-vitro system can recapitulate most of the changes during the period of brain development making it an ideal model for predicting neurotoxic effects. In the present study, we verified the possible DNT of epoxomicin which is a naturally occurring selective proteasome inhibitor with anti-inflammatory activity. Rat neural progenitor cells were isolated from rat embryos (E14) extracted from placental tissue. The cortices were aseptically dissected out from the brains of the fetuses and the tissues were triturated by repeated passage through a fire-polished constricted Pasteur pipette. The dispersed tissues were allowed to settle for 3 min. The supernatant was, then, transferred to a fresh tube and centrifuged at 1,000 g for 5 min. The pellet was placed in Hank’s balanced salt solution cultured as free-floating neurospheres in proliferation medium. Two doses of epoxomicin (1µM and 10µM) were used in cultured neuropsheres for a period of 14 days. For proliferation analysis, spheres were cultured in proliferation medium. After 0, 4, 5, 11, and 14 days, sphere size was determined by software analyses. The diameter of each neurosphere was measured and exported to excel file further to statistical analysis. For viability analysis, trypsin-EDTA solution were added to neurospheres for 3 min to dissociate them into single cells suspension, then viability evaluated by the Trypan Blue exclusion test. Epoxomicin was found to affect proliferation and viability of neuropsheres, these effects were positively correlated to doses and progress of time. This study confirms the DNT effects of epoxomicin on 3D neurospheres model. The effects on proliferation suggest possible gross morphologic changes while the decrease in viability propose possible focal lesion on exposure to epoxomicin during early childhood.

Keywords: neural progentor cells, epoxomicin, neurosphere, medical and health sciences

Procedia PDF Downloads 428

Authors: Neveen AlQasas, Daniel Johnson

Abstract:

Membranes for water treatment are an established technology that attracts great attention due to its simplicity and cost effectiveness. However, membranes in operation suffer from the adverse effect of membrane fouling. Bio-fouling is a phenomenon that occurs at the water-membrane interface, and is a dynamic process that is initiated by the adsorption of dissolved organic material, including biomacromolecules, on the membrane surface. After initiation, attachment of microorganisms occurs, followed by biofilm growth. The biofilm blocks the pores of the membrane and consequently results in reducing the water flux. Moreover, the presence of a fouling layer can have a substantial impact on the membrane separation properties. Understanding the mechanism of the initiation phase of biofouling is a key point in eliminating the biofouling on membrane surfaces. The adhesion and attachment of different fouling materials is affected by the surface properties of the membrane materials. Therefore, surface properties of different polymeric materials had been studied in terms of their surface energies and Hansen solubility parameters (HSP). The difference between the combined HSP parameters (HSP distance) allows prediction of the affinity of two materials to each other. The possibilities of measuring the HSP of different polymer films via surface measurements, such as contact angle has been thoroughly investigated. Knowing the HSP of a membrane material and the HSP of a specific foulant, facilitate the estimation of the HSP distance between the two, and therefore the strength of attachment to the surface. Contact angle measurements using fourteen different solvents on five different polymeric films were carried out using the sessile drop method. Solvents were ranked as good or bad solvents using different ranking method and ranking was used to calculate the HSP of each polymeric film. Results clearly indicate the absence of a direct relation between contact angle values of each film and the HSP distance between each polymer film and the solvents used. Therefore, estimating HSP via contact angle alone is not sufficient. However, it was found if the surface tensions and viscosities of the used solvents are taken in to the account in the analysis of the contact angle values, a prediction of the HSP from contact angle measurements is possible. This was carried out via training of a neural network model. The trained neural network model has three inputs, contact angle value, surface tension and viscosity of solvent used. The model is able to predict the HSP distance between the used solvent and the tested polymer (material). The HSP distance prediction is further used to estimate the total and individual HSP parameters of each tested material. The results showed an accuracy of about 90% for all the five studied films

Keywords: surface characterization, hansen solubility parameter estimation, contact angle measurements, artificial neural network model, surface measurements

Procedia PDF Downloads 94

Authors: Mario Isaza-Ruget, Claudia C. Colmenares-Mejia, Nancy Yomayusa, Camilo A. González, Andres Cely, Jossie Murcia

Abstract:

Quantification of risks associated with complications development from chronic kidney disease (CKD) through accurate survival models can help with patient management. A retrospective cohort that included patients diagnosed with CKD from a primary care program and followed up between 2013 and 2018 was carried out. Time-dependent and static covariates associated with demographic, clinical, and laboratory factors were included. Deep Learning (DL) survival analyzes were developed for three CKD outcomes: CKD stage progression, >25% decrease in Estimated Glomerular Filtration Rate (eGFR), and Renal Replacement Therapy (RRT). Models were evaluated and compared with Random Survival Forest (RSF) based on concordance index (C-index) metric. 2.143 patients were included. Two models were developed for each outcome, Deep Neural Network (DNN) model reported C-index=0.9867 for CKD stage progression; C-index=0.9905 for reduction in eGFR; C-index=0.9867 for RRT. Regarding the RSF model, C-index=0.6650 was reached for CKD stage progression; decreased eGFR C-index=0.6759; RRT C-index=0.8926. DNN models applied in survival analysis context with considerations of longitudinal covariates at the start of follow-up can predict renal stage progression, a significant decrease in eGFR and RRT. The success of these survival models lies in the appropriate definition of survival times and the analysis of covariates, especially those that vary over time.

Keywords: artificial intelligence, chronic kidney disease, deep neural networks, survival analysis

Procedia PDF Downloads 136

Authors: Yitao Lei, Xingxiang Zhai, Burra Venkata Durga Kumar

Abstract:

As the quantity and complexity of computing in large-scale software systems increase, distributed system computing becomes increasingly important. The distributed system realizes high-performance computing by collaboration between different computing resources. If there are no efficient resource scheduling resources, the abuse of distributed computing may cause resource waste and high costs. However, resource scheduling is usually an NP-hard problem, so we cannot find a general solution. However, some optimization algorithms exist like genetic algorithm, ant colony optimization, etc. The large scale of distributed systems makes this traditional optimization algorithm challenging to work with. Heuristic and machine learning algorithms are usually applied in this situation to ease the computing load. As a result, we do a review of traditional resource scheduling optimization algorithms and try to introduce a deep reinforcement learning method that utilizes the perceptual ability of neural networks and the decision-making ability of reinforcement learning. Using the machine learning method, we try to find important factors that influence the performance of distributed system computing and help the distributed system do an efficient computing resource scheduling. This paper surveys the application of deep reinforcement learning on distributed system computing resource scheduling proposes a deep reinforcement learning method that uses a recurrent neural network to optimize the resource scheduling, and proposes the challenges and improvement directions for DRL-based resource scheduling algorithms.

Keywords: resource scheduling, deep reinforcement learning, distributed system, artificial intelligence

Procedia PDF Downloads 113

Authors: Simon Espigolé, Igor Shalyminov, Helen Hastie

Abstract:

Deep machine learning techniques including deep neural networks (DNN) have been used to model language and dialogue for conversational agents to perform tasks, such as giving technical support and also for general chit-chat. They have been shown to be capable of generating long, diverse and coherent sentences in end-to-end dialogue systems and natural language generation. However, these systems tend to imitate the training data and will only generate the concepts and language within the scope of what they have been trained on. This work explores how deep neural networks can be used in a task that would normally require human creativity, whereby the human would read the movie description and/or watch the movie and come up with a compelling, interesting movie title. This task differs from simple summarization in that the movie title may not necessarily be derivable from the content or semantics of the movie description. Here, we train a type of DNN called a sequence-to-sequence model (seq2seq) that takes as input a short textual movie description and some information on e.g. genre of the movie. It then learns to output a movie title. The idea is that the DNN will learn certain techniques and approaches that the human movie titler may deploy that may not be immediately obvious to the human-eye. To give an example of a generated movie title, for the movie synopsis: ‘A hitman concludes his legacy with one more job, only to discover he may be the one getting hit.’; the original, true title is ‘The Driver’ and the one generated by the model is ‘The Masquerade’. A human evaluation was conducted where the DNN output was compared to the true human-generated title, as well as a number of baselines, on three 5-point Likert scales: ‘creativity’, ‘naturalness’ and ‘suitability’. Subjects were also asked which of the two systems they preferred. The scores of the DNN model were comparable to the scores of the human-generated movie title, with means m=3.11, m=3.12, respectively. There is room for improvement in these models as they were rated significantly less ‘natural’ and ‘suitable’ when compared to the human title. In addition, the human-generated title was preferred overall 58% of the time when pitted against the DNN model. These results, however, are encouraging given the comparison with a highly-considered, well-crafted human-generated movie title. Movie titles go through a rigorous process of assessment by experts and focus groups, who have watched the movie. This process is in place due to the large amount of money at stake and the importance of creating an effective title that captures the audiences’ attention. Our work shows progress towards automating this process, which in turn may lead to a better understanding of creativity itself.

Keywords: creativity, deep machine learning, natural language generation, movies

Procedia PDF Downloads 327

Authors: Virender Singh, Mathew Rees, Simon Hampton, Sivaram Annadurai

Abstract:

Plant identification is a challenging task that aims to identify the family, genus, and species according to plant morphological features. Automated deep learning-based computer vision algorithms are widely used for identifying plants and can help users narrow down the possibilities. However, numerous morphological similarities between and within species render correct classification difficult. In this paper, we tested custom convolution neural network (CNN) and vision transformer (ViT) based models using the PyTorch framework to classify plants. We used a large dataset of 88,000 provided by the Royal Horticultural Society (RHS) and a smaller dataset of 16,000 images from the PlantClef 2015 dataset for classifying plants at genus and species levels, respectively. Our results show that for classifying plants at the genus level, ViT models perform better compared to CNN-based models ResNet50 and ResNet-RS-420 and other state-of-the-art CNN-based models suggested in previous studies on a similar dataset. ViT model achieved top accuracy of 83.3% for classifying plants at the genus level. For classifying plants at the species level, ViT models perform better compared to CNN-based models ResNet50 and ResNet-RS-420, with a top accuracy of 92.5%. We show that the correct set of augmentation techniques plays an important role in classification success. In conclusion, these results could help end users, professionals and the general public alike in identifying plants quicker and with improved accuracy.

Keywords: plant identification, CNN, image processing, vision transformer, classification

Procedia PDF Downloads 105

Authors: Abdul Rehman, Bo Liu

Abstract:

Axial flow turbines are commonly designed with high loads that generate strong secondary flows and result in high secondary losses. These losses contribute to almost 30% to 50% of the total losses. Non-axisymmetric endwall profiling is one of the passive control technique to reduce the secondary flow loss. In this paper, the non-axisymmetric endwall profile construction and optimization for the stator endwalls are presented to improve the efficiency of a high pressure turbine. The commercial code NUMECA Fine/ Design3D coupled with Fine/Turbo was used for the numerical investigation, design of experiments and the optimization. All the flow simulations were conducted by using steady RANS and Spalart-Allmaras as a turbulence model. The non-axisymmetric endwalls of stator hub and shroud were created by using the perturbation law based on Bezier Curves. Each cut having multiple control points was supposed to be created along the virtual streamlines in the blade channel. For the design of experiments, each sample was arbitrarily generated based on values automatically chosen for the control points defined during parameterization. The Optimization was achieved by using two algorithms i.e. the stochastic algorithm and gradient-based algorithm. For the stochastic algorithm, a genetic algorithm based on the artificial neural network was used as an optimization method in order to achieve the global optimum. The evaluation of the successive design iterations was performed using artificial neural network prior to the flow solver. For the second case, the conjugate gradient algorithm with a three dimensional CFD flow solver was used to systematically vary a free-form parameterization of the endwall. This method is efficient and less time to consume as it requires derivative information of the objective function. The objective function was to maximize the isentropic efficiency of the turbine by keeping the mass flow rate as constant. The performance was quantified by using a multi-objective function. Other than these two classifications of the optimization methods, there were four optimizations cases i.e. the hub only, the shroud only, and the combination of hub and shroud. For the fourth case, the shroud endwall was optimized by using the optimized hub endwall geometry. The hub optimization resulted in an increase in the efficiency due to more homogenous inlet conditions for the rotor. The adverse pressure gradient was reduced but the total pressure loss in the vicinity of the hub was increased. The shroud optimization resulted in an increase in efficiency, total pressure loss and entropy were reduced. The combination of hub and shroud did not show overwhelming results which were achieved for the individual cases of the hub and the shroud. This may be caused by fact that there were too many control variables. The fourth case of optimization showed the best result because optimized hub was used as an initial geometry to optimize the shroud. The efficiency was increased more than the individual cases of optimization with a mass flow rate equal to the baseline design of the turbine. The results of artificial neural network and conjugate gradient method were compared.

Keywords: artificial neural network, axial turbine, conjugate gradient method, non-axisymmetric endwall, optimization

Procedia PDF Downloads 226

Authors: Adriana Martínez-Arias, John Josephraj Selvaraj, Luis Octavio González-Salcedo

Abstract:

In the Colombian Pacific, Dorado fish (Coryphaena hippurus L.) fisheries is of great commercial interest. However, its habitat and fisheries may be affected by climatic change especially by the actual increase in sea surface temperature. Hence, it is of interest to study the dynamics of these species fishing zones. In this study, we developed Artificial Neural Networks (ANN) models to predict Catch per Unit Effort (CPUE) as an indicator of species abundance. The model was based on four oceanographic variables (Chlorophyll a, Sea Surface Temperature, Sea Level Anomaly and Bathymetry) derived from satellite data. CPUE datasets for model training and cross-validation were obtained from logbooks of commercial fishing vessel. Sea surface Temperature for Colombian Pacific were projected under Representative Concentration Pathway (RCP) scenarios 4.5 and 8.5 using Coupled Model Intercomparison Project Phase 5 (CMIP5) and CPUE maps were created. Our results indicated that an increase in sea surface temperature reduces the potential fishing zones of this species in the Colombian Pacific. We conclude that ANN is a reliable tool for simulation of climate change effects on the potential fishing zones. This research opens a future agenda for other species that have been affected by climate change.

Keywords: climatic change, artificial neural networks, dorado fish, CPUE

Procedia PDF Downloads 246

Authors: Yu-ding Du, Qi-lian Bao, Nassim Bessaad, Lin Liu

Abstract:

The Global Navigation Satellite System (GNSS) is regarded as an effective approach for the purpose of replacing the large amount used track-side balises in modern train localization systems. This paper describes a method based on the data fusion of a GNSS receiver sensor and an odometer sensor that can significantly improve the positioning accuracy. A digital track map is needed as another sensor to project two-dimensional GNSS position to one-dimensional along-track distance due to the fact that the train’s position can only be constrained on the track. A model trained by BP neural network is used to estimate the trend positioning error which is related to the specific location and proximate processing of the digital track map. Considering that in some conditions the satellite signal failure will lead to the increase of GNSS positioning error, a detection step for GNSS signal is applied. An adaptive weighted fusion algorithm is presented to reduce the standard deviation of train speed measurement. Finally an Extended Kalman Filter (EKF) is used for the fusion of the projected 1-D GNSS positioning data and the 1-D train speed data to get the estimate position. Experimental results suggest that the proposed method performs well, which can reduce positioning error notably.

Keywords: multi-sensor data fusion, train positioning, GNSS, odometer, digital track map, map matching, BP neural network, adaptive weighted fusion, Kalman filter

Procedia PDF Downloads 254

Authors: Milica Karadžić, Lidija Jevrić, Sanja Podunavac-Kuzmanović, Strahinja Kovačević, Anamarija Mandić, Aleksandar Oklješa, Andrea Nikolić, Marija Sakač, Katarina Penov Gaši

Abstract:

Using chemometric approach, the relationships between the chromatographic lipophilicity and in silico molecular descriptors for twenty-nine selected steroid derivatives were studied. The chromatographic lipophilicity was predicted using artificial neural networks (ANNs) method. The most important in silico molecular descriptors were selected applying stepwise selection (SS) paired with partial least squares (PLS) method. Molecular descriptors with satisfactory variable importance in projection (VIP) values were selected for ANN modeling. The usefulness of generated models was confirmed by detailed statistical validation. High agreement between experimental and predicted values indicated that obtained models have good quality and high predictive ability. Global sensitivity analysis (GSA) confirmed the importance of each molecular descriptor used as an input variable. High-quality networks indicate a strong non-linear relationship between chromatographic lipophilicity and used in silico molecular descriptors. Applying selected molecular descriptors and generated ANNs the good prediction of chromatographic lipophilicity of the studied steroid derivatives can be obtained. This article is based upon work from COST Actions (CM1306 and CA15222), supported by COST (European Cooperation and Science and Technology).

Keywords: artificial neural networks, chemometrics, global sensitivity analysis, liquid chromatography, steroids

Procedia PDF Downloads 347

Authors: Tim Steinhaus, Christian Beidl

Abstract:

Urban air quality and climate protection are two major challenges for future mobility systems. Despite the steady reduction of pollutant emissions from vehicles over past decades, local immission load within cities partially still reaches heights, which are considered hazardous to human health. Although traffic-related emissions account for a major part of the overall urban pollution, modeling the exact interaction remains challenging. In this paper, a novel approach for the determination of the emission-immission interaction on the basis of neural network modeling for traffic induced NO2-immission load within a near-traffic hot-spot scenario is presented. In a detailed sensitivity analysis, the significance of relevant influencing variables on the prevailing NO2 concentration is initially analyzed. Based on this, the generation process of the model is described, in which not only environmental influences but also the vehicle fleet composition including its associated segment- and certification-specific real driving emission factors are derived and used as input quantities. The validity of this approach, which has been presented in the past, is re-examined in this paper using updated data on vehicle emissions and recent immission measurement data. Within the framework of a final scenario analysis, the future development of the immission load is forecast for different developments in the vehicle fleet composition. It is shown that immission levels of less than half of today’s yearly average limit values are technically feasible in hot-spot situations.

Keywords: air quality, emission, emission-immission-interaction, immission, NO2, zero impact

Procedia PDF Downloads 127

Authors: Parker Murray, Marci Johnson, Tyson S. Burnham, Alina K. Fong, Mark D. Allen, Bruce McIff

Abstract:

Purpose: Pathological dysregulation of Neurovascular Coupling (NVC) caused by mild traumatic brain injury (mTBI) is the predominant source of chronic post-concussion syndrome (PCS) symptomology. fNCI has the ability to localize dysregulation in NVC by measuring blood-oxygen-level-dependent (BOLD) signaling during the performance of fMRI-adapted neuropsychological evaluations. With fNCI, 57 brain areas consistently affected by concussion were identified as PCS neural markers, which were validated on large samples of concussion patients and healthy controls. These neuromarkers provide the basis for a computation of PCS severity which is referred to as the Severity Index Score (SIS). The SIS has proven valuable in making pre-treatment decisions, monitoring treatment efficiency, and assessing long-term stability of outcomes. Methods and Materials: After being scanned while performing various cognitive tasks, 476 concussed patients received an SIS score based on the neural dysregulation of the 57 previously identified brain regions. These scans provide an objective measurement of attentional, subcortical, visual processing, language processing, and executive functioning abilities, which were used as biomarkers for post-concussive neural dysregulation. Initial SIS scores were used to develop individualized therapy incorporating cognitive, occupational, and neuromuscular modalities. These scores were also used to establish pre-treatment benchmarks and measure post-treatment improvement. Results: Changes in SIS were calculated in percent change from pre- to post-treatment. Patients showed a mean improvement of 76.5 percent (σ= 23.3), and 75.7 percent of patients showed at least 60 percent improvement. Longitudinal reassessment of 24 of the patients, measured an average of 7.6 months post-treatment, shows that SIS improvement is maintained and improved, with an average of 90.6 percent improvement from their original scan. Conclusions: fNCI provides a reliable measurement of NVC allowing for identification of concussion pathology. Additionally, fNCI derived SIS scores direct tailored therapy to restore NVC, subsequently resolving chronic PCS resulting from mTBI.

Keywords: concussion, functional magnetic resonance imaging (fMRI), neurovascular coupling (NVC), post-concussion syndrome (PCS)

Procedia PDF Downloads 360

Authors: Shanshan Zhu, Mohammad Nasim

Abstract:

Financial fraud continues to be a major threat to financial institutions across the world, causing colossal money losses and undermining public trust. Fraud prevention techniques, based on hard rules, have become ineffective due to evolving patterns of fraud in recent times. Against such a background, the present study probes into distinct methodologies that exploit emergent AI-driven techniques to further strengthen fraud detection. We would like to compare the performance of generative adversarial networks and graph neural networks with other popular techniques, like gradient boosting, random forests, and neural networks. To this end, we would recommend integrating all these state-of-the-art models into one robust, flexible, and smart system for real-time anomaly and fraud detection. To overcome the challenge, we designed synthetic data and then conducted pattern recognition and unsupervised and supervised learning analyses on the transaction data to identify which activities were fishy. With the use of actual financial statistics, we compare the performance of our model in accuracy, speed, and adaptability versus conventional models. The results of this study illustrate a strong signal and need to integrate state-of-the-art, AI-driven fraud detection solutions into frameworks that are highly relevant to the financial domain. It alerts one to the great urgency that banks and related financial institutions must rapidly implement these most advanced technologies to continue to have a high level of security.

Keywords: AI-driven fraud detection, financial security, machine learning, anomaly detection, real-time fraud detection

Procedia PDF Downloads 44

Authors: Atitallah Sofien, Bouyahia Olfa, Romdhani Meriam, Missaoui Nada, Ben Rabeh Rania, Yahyaoui Salem, Mazigh Sonia, Boukthir Samir

Abstract:

Introduction: Spondylothoracic dysostosis, also known as Jarcho-Levin syndrome, is defined by a shortened neck and thorax, a protruding abdomen, inguinal and umbilical hernias, atypical spinal structure and rib fusion, leading to restricted chest movement or difficulty in breathing, along with urinary tract abnormalities and, potentially severe scoliosis. Aim: This is the case of a patient diagnosed with Jarcho-Levin syndrome, aiming to detail the range of abnormalities observed in this syndrome, the observed complications, and the therapeutic approaches employed. Results: A three-month-old male infant, born of a consanguineous marriage, delivered at full term by cesarean section, was admitted to the pediatric department for severe acute bronchiolitis. In his prenatal history, morphological ultrasound revealed macrosomia, a shortened spine, irregular vertebrae with thickened skin, normal fetal cardiac ultrasound, and the absence of the right kidney. His perinatal history included respiratory distress, requiring ventilatory support for five days. Upon physical examination, he had stunted growth, scoliosis, a short neck and trunk, longer upper limbs compared to lower limbs, varus equinus in the right foot, a neural tube defect, a low hairline, and low-set ears. Spondylothoracic dysostosis was suspected, leading to further investigations, including a normal transfontaneous ultrasound, a spinal cord ultrasound revealing a lipomyelocele-type closed dysraphism with a low-attached cord, an abdominal ultrasound indicating a single left kidney, and a cardiac ultrasound identifying Kommerell syndrome. Due to a lack of resources, genetic testing could not be performed, and the diagnosis was based on clinical criteria. Conclusion: Jarcho-Levin syndrome can result in a mortality rate of about 50%, primarily due to respiratory complications associated with thoracic insufficiency syndrome. Other complications, like heart and neural tube defects, can also lead to premature mortality. Therefore, early diagnosis and comprehensive treatment involving various specialists are essential.

Keywords: Jarcho-Levin syndrome, congenital disorder, scoliosis, spondylothoracic dysostosis, neural tube defect

Procedia PDF Downloads 58

Authors: Rao Fu, Jie Chen, Shutian Zeng, Yiping Zhuang, Agus Sudjianto

Abstract:

Generative Adversarial Net (GAN) has proved to be a powerful machine learning tool in image data analysis and generation. In this paper, we propose to use Conditional Generative Adversarial Net (CGAN) to learn and simulate time series data. The conditions include both categorical and continuous variables with different auxiliary information. Our simulation studies show that CGAN has the capability to learn different types of normal and heavy-tailed distributions, as well as dependent structures of different time series. It also has the capability to generate conditional predictive distributions consistent with training data distributions. We also provide an in-depth discussion on the rationale behind GAN and the neural networks as hierarchical splines to establish a clear connection with existing statistical methods of distribution generation. In practice, CGAN has a wide range of applications in market risk and counterparty risk analysis: it can be applied to learn historical data and generate scenarios for the calculation of Value-at-Risk (VaR) and Expected Shortfall (ES), and it can also predict the movement of the market risk factors. We present a real data analysis including a backtesting to demonstrate that CGAN can outperform Historical Simulation (HS), a popular method in market risk analysis to calculate VaR. CGAN can also be applied in economic time series modeling and forecasting. In this regard, we have included an example of hypothetical shock analysis for economic models and the generation of potential CCAR scenarios by CGAN at the end of the paper.

Keywords: conditional generative adversarial net, market and credit risk management, neural network, time series

Procedia PDF Downloads 144

Authors: Farhad Salek, Shahaboddin Resalati

Abstract:

The production of electric vehicles (EVs) featuring lithium-ion battery technology has substantially escalated over the past decade, demonstrating a steady and persistent upward trajectory. The imminent retirement of electric vehicle (EV) batteries after approximately eight years underscores the critical need for their redirection towards recycling, a task complicated by the current inadequacy of recycling infrastructures globally. A potential solution for such concerns involves extending the operational lifespan of electric vehicle (EV) batteries through their utilization in stationary energy storage systems during secondary applications. Such adoptions, however, require addressing the safety concerns associated with batteries’ knee points and thermal runaways. This paper develops an accurate mathematical model representative of the second-life battery packs from a cell-to-pack scale using an equivalent circuit model (ECM) methodology. Neural network algorithms are employed to forecast the degradation parameters based on the EV batteries' aging history to develop a degradation model. The degradation model is integrated with the ECM to reflect the impacts of the cycle aging mechanism on battery parameters during operation. The developed model is tested under real-life load profiles to evaluate the life span of the batteries in various operating conditions. The methodology and the algorithms introduced in this paper can be considered the basis for Battery Management System (BMS) design and techno-economic analysis of such technologies.

Keywords: second life battery, electric vehicles, degradation, neural network

Procedia PDF Downloads 66

Authors: Lizhi Ma, Dan Liu

Abstract:

Pipeline parallelism has been widely used to accelerate distributed deep learning to alleviate GPU memory bottlenecks and to ensure that models can be trained and deployed smoothly under limited graphics memory conditions. However, in highly heterogeneous distributed clusters, traditional model partitioning methods are not able to achieve load balancing. The overlap of communication and computation is also a big challenge. In this paper, HePipe is proposed, an efficient pipeline parallel training method for highly heterogeneous clusters. According to the characteristics of the neural network model pipeline training task, oriented to the 2-level heterogeneous cluster computing topology, a training method based on the 2-level stage division of neural network modeling and partitioning is designed to improve the parallelism. Additionally, a multi-forward 1F1B scheduling strategy is designed to accelerate the training time of each stage by executing the computation units in advance to maximize the overlap between the forward propagation communication and backward propagation computation. Finally, a dynamic recomputation strategy based on task memory requirement prediction is proposed to improve the fitness ratio of task and memory, which improves the throughput of the cluster and solves the memory shortfall problem caused by memory differences in heterogeneous clusters. The empirical results show that HePipe improves the training speed by 1.6×−2.2× over the existing asynchronous pipeline baselines.

Keywords: pipeline parallelism, heterogeneous cluster, model training, 2-level stage partitioning

Procedia PDF Downloads 20

Authors: Devadrita Dey Sarkar

Abstract:

Regardless of the many technologic advances in the past decade, increased training and experience, and the obvious benefits of uniform standards, the false-negative rate in screening mammography remains unacceptably high .A computer aided neural network classification of regions of suspicion (ROS) on digitized mammograms is presented in this abstract which employs features extracted by a new technique based on independent component analysis. CAD is a concept established by taking into account equally the roles of physicians and computers, whereas automated computer diagnosis is a concept based on computer algorithms only. With CAD, the performance by computers does not have to be comparable to or better than that by physicians, but needs to be complementary to that by physicians. In fact, a large number of CAD systems have been employed for assisting physicians in the early detection of breast cancers on mammograms. A CAD scheme that makes use of lateral breast images has the potential to improve the overall performance in the detection of breast lumps. Because breast lumps can be detected reliably by computer on lateral breast mammographs, radiologists’ accuracy in the detection of breast lumps would be improved by the use of CAD, and thus early diagnosis of breast cancer would become possible. In the future, many CAD schemes could be assembled as packages and implemented as a part of PACS. For example, the package for breast CAD may include the computerized detection of breast nodules, as well as the computerized classification of benign and malignant nodules. In order to assist in the differential diagnosis, it would be possible to search for and retrieve images (or lesions) with these CAD systems, which would be reliable and useful method for quantifying the similarity of a pair of images for visual comparison by radiologists.

Keywords: CAD(computer-aided design), lesions, neural network, ROS(region of suspicion)

Procedia PDF Downloads 456

Authors: Maria Pavlova

Abstract:

In our days, it is possible to put the camera on different vehicles like quadcopter, train, airplane and etc. The camera also can be the input sensor in many different systems. That means the object recognition like non separate part of monitoring control can be key part of the most intelligent systems. The aim of this paper is to focus of the object recognition process during vehicles movement. During the vehicle’s movement the camera takes pictures from the environment without storage in Data Base. In case the camera detects a special object (for example human or animal), the system saves the picture and sends it to the work station in real time. This functionality will be very useful in emergency or security situations where is necessary to find a specific object. In another application, the camera can be mounted on crossroad where do not have many people and if one or more persons come on the road, the traffic lights became the green and they can cross the road. In this papers is presented the system has solved the aforementioned problems. It is presented architecture of the object recognition system includes the camera, Raspberry platform, GPS system, neural network, software and Data Base. The camera in the system takes the pictures. The object recognition is done in real time using the OpenCV library and Raspberry microcontroller. An additional feature of this library is the ability to display the GPS coordinates of the captured objects position. The results from this processes will be sent to remote station. So, in this case, we can know the location of the specific object. By neural network, we can learn the module to solve the problems using incoming data and to be part in bigger intelligent system. The present paper focuses on the design and integration of the image recognition like a part of smart systems.

Keywords: camera, object recognition, OpenCV, Raspberry

Procedia PDF Downloads 219

Authors: Dhruv J. Limaye, Hanga Galfalvy, Cheick A. Sissoko, Yung-yu Huang, Chunanning Tang, Ying Liu, Shu-Chi Hsiung, Andrew J. Dwork, Gorazd B. Rosoklija, Victoria Arango, Lewis Brown, J. John Mann, Maura Boldrini

Abstract:

Memory and emotion require hippocampal cell viability and connectivity and are disrupted in major depressive disorder (MDD). Applying shotgun proteomics and stereological quantification of neural progenitor cells (NPCs), intermediate neural progenitors (INPs), and mature granule neurons (GNs), to postmortem human hippocampus, identified differentially expressed proteins (DEPs), and fewer NPCs, INPs and GNs, in untreated MDD (uMDD) compared with non-psychiatric controls (CTRL) and antidepressant-treated MDD (MDDT). DEPs lower in uMDD vs. CTRL promote mitosis, differentiation, and prevent apoptosis. DEPs higher in uMDD vs. CTRL inhibit the cell cycle, and regulate cell adhesion, neurite outgrowth, and DNA repair. DEPs lower in MDDT vs. uMDD block cell proliferation. We observe group-specific correlations between numbers of NPCs, INPs, and GNs and an abundance of proteins regulating mitosis, differentiation, and apoptosis. Altered protein expression underlies hippocampus cellular and volume loss in uMDD, supports a trophic effect of antidepressants, and offers new treatment targets.

Keywords: proteomics, hippocampus, depression, mitosis, migration, differentiation, mitochondria, apoptosis, antidepressants, human brain

Procedia PDF Downloads 101

Authors: McClain Thiel

Abstract:

Object detection and localization has historically required two or more sensors due to the loss of information from 3D to 2D space, however, most surveillance systems currently in use in the real world only have one sensor per location. Generally, this consists of a single low-resolution camera positioned above the area under observation (mall, jewelry store, traffic camera). This is not sufficient for robust 3D tracking for applications such as security or more recent relevance, contract tracing. This paper proposes a lightweight system for 3D person tracking that requires no additional hardware, based on compressed object detection convolutional-nets, facial landmark detection, and projective geometry. This approach involves classifying the target into a demographic category and then making assumptions about the relative locations of facial landmarks from the demographic information, and from there using simple projective geometry and known constants to find the target's location in 3D space. Preliminary testing, although severely lacking, suggests reasonable success in 3D tracking under ideal conditions.

Keywords: monocular distancing, computer vision, facial analysis, 3D localization

Procedia PDF Downloads 142

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 199

Authors: Xi Liu, Fan Yuan

Abstract:

In the evolving landscape of physics-based machine learning (PBML), particularly within fluid dynamics and its applications in electromechanical engineering, robot vision, and robot learning, achieving precision and alignment with researchers' specific needs presents a formidable challenge. In response, this work proposes a methodology that integrates neural transformation with a modified smoothed particle hydrodynamics model for generating transformed 3D fluid simulations. This approach is useful for nanoscale science, where the unique and complex behaviors of viscoelastic medium demand accurate neurally-transformed simulations for materials understanding and manipulation. In electromechanical engineering, the method enhances the design and functionality of fluid-operated systems, particularly microfluidic devices, contributing to advancements in nanomaterial design, drug delivery systems, and more. The proposed approach also aligns with the principles of PBML, offering advantages such as multi-fluid stylization and consistent particle attribute transfer. This capability is valuable in various fields where the interaction of multiple fluid components is significant. Moreover, the application of neurally-transformed hydrodynamical models extends to manufacturing processes, such as the production of microelectromechanical systems, enhancing efficiency and cost-effectiveness. The system's ability to perform neural transfer on 3D fluid scenes using a deep learning algorithm alongside physical models further adds a layer of flexibility, allowing researchers to tailor simulations to specific needs across scientific and engineering disciplines.

Keywords: physics-based machine learning, robot vision, robot learning, hydrodynamics

Procedia PDF Downloads 66

Authors: Shanaka Gunasekara, Maleen Jayasuriya, Nalin Harischandra, Lilantha Samaranayake, Gamini Dissanayake

Abstract:

One serious facet of the worsening Human-Elephant conflict (HEC) in nations such as Sri Lanka involves elephant-train collisions. Endangered Asian elephants are maimed or killed during such accidents, which also often result in orphaned or disabled elephants, contributing to the phenomenon of lone elephants. These lone elephants are found to be more likely to attack villages and showcase aggressive behaviour, which further exacerbates the overall HEC. Furthermore, Railway Services incur significant financial losses and disruptions to services annually due to such accidents. Most elephant-train collisions occur due to a lack of adequate reaction time. This is due to the significant stopping distance requirements of trains, as the full braking force needs to be avoided to minimise the risk of derailment. Thus, poor driver visibility at sharp turns, nighttime operation, and poor weather conditions are often contributing factors to this problem. Initial investigations also indicate that most collisions occur in localised “hotspots” where elephant pathways/corridors intersect with railway tracks that border grazing land and watering holes. Taking these factors into consideration, this work proposes the leveraging of recent developments in Convolutional Neural Network (CNN) technology to detect elephants using an RGB/infrared capable camera around known hotspots along the railway track. The CNN was trained using a curated dataset of elephants collected on field visits to elephant sanctuaries and wildlife parks in Sri Lanka. With this vision-based detection system at its core, a prototype unit of an early warning system was designed and tested. This weatherised and waterproofed unit consists of a Reolink security camera which provides a wide field of view and range, an Nvidia Jetson Xavier computing unit, a rechargeable battery, and a solar panel for self-sufficient functioning. The prototype unit was designed to be a low-cost, low-power and small footprint device that can be mounted on infrastructures such as poles or trees. If an elephant is detected, an early warning message is communicated to the train driver using the GSM network. A mobile app for this purpose was also designed to ensure that the warning is clearly communicated. A centralized control station manages and communicates all information through the train station network to ensure coordination among important stakeholders. Initial results indicate that detection accuracy is sufficient under varying lighting situations, provided comprehensive training datasets that represent a wide range of challenging conditions are available. The overall hardware prototype was shown to be robust and reliable. We envision a network of such units may help contribute to reducing the problem of elephant-train collisions and has the potential to act as an important surveillance mechanism in dealing with the broader issue of human-elephant conflicts.

Keywords: computer vision, deep learning, human-elephant conflict, wildlife early warning technology

Procedia PDF Downloads 226

Authors: Amanda Kavner, Richard Lamb

Abstract:

Faster growth in science and technology of other nations may make staying globally competitive more difficult without shifting focus on how science is taught in US classes. An integral part of learning science involves visual and spatial thinking since complex, and real-world phenomena are often expressed in visual, symbolic, and concrete modes. The primary barrier to spatial thinking and visual literacy in Science, Technology, Engineering, and Math (STEM) fields is representational competence, which includes the ability to generate, transform, analyze and explain representations, as opposed to generic spatial ability. Although the relationship is known between the foundational visual literacy and the domain-specific science literacy, science literacy as a function of science learning is still not well understood. Moreover, the need for a more reliable measure is necessary to design resources which enhance the fundamental visuospatial cognitive processes behind scientific literacy. To support the improvement of students’ representational competence, first visualization skills necessary to process these science representations needed to be identified, which necessitates the development of an instrument to quantitatively measure visual literacy. With such a measure, schools, teachers, and curriculum designers can target the individual skills necessary to improve students’ visual literacy, thereby increasing science achievement. This project details the development of an artificial neural network capable of measuring science literacy using functional Near-Infrared Spectroscopy (fNIR) data. This data was previously collected by Project LENS standing for Leveraging Expertise in Neurotechnologies, a Science of Learning Collaborative Network (SL-CN) of scholars of STEM Education from three US universities (NSF award 1540888), utilizing mental rotation tasks, to assess student visual literacy. Hemodynamic response data from fNIRsoft was exported as an Excel file, with 80 of both 2D Wedge and Dash models (dash) and 3D Stick and Ball models (BL). Complexity data were in an Excel workbook separated by the participant (ID), containing information for both types of tasks. After changing strings to numbers for analysis, spreadsheets with measurement data and complexity data were uploaded to RapidMiner’s TurboPrep and merged. Using RapidMiner Studio, a Gradient Boosted Trees artificial neural network (ANN) consisting of 140 trees with a maximum depth of 7 branches was developed, and 99.7% of the ANN predictions are accurate. The ANN determined the biggest predictors to a successful mental rotation are the individual problem number, the response time and fNIR optode #16, located along the right prefrontal cortex important in processing visuospatial working memory and episodic memory retrieval; both vital for science literacy. With an unbiased measurement of science literacy provided by psychophysiological measurements with an ANN for analysis, educators and curriculum designers will be able to create targeted classroom resources to help improve student visuospatial literacy, therefore improving science literacy.

Keywords: artificial intelligence, artificial neural network, machine learning, science literacy, neuroscience

Procedia PDF Downloads 121

Authors: Delphine Ribes Lemay, Valentine Bernasconi, André Andrade, Lara DéFayes, Mathieu Salzmann, FréDéRic Kaplan, Nicolas Henchoz

Abstract:

Digitization and preservation of large heritage induce high maintenance costs to keep up with the technical standards and ensure sustainable access. Creating impactful usage is instrumental to justify the resources for long-term preservation. The Museum für Gestaltung of Zurich holds one of the biggest poster collections of the world from which 52’000 were digitised. In the process of building a digital installation to valorize the collection, one objective was to develop an algorithm capable of predicting the next poster to show according to the ones already displayed. The work presented here describes the steps to build an algorithm able to automatically create sequences of posters reflecting associations performed by curator and professional designers. The exposed challenge finds similarities with the domain of song playlist algorithms. Recently, artificial intelligence techniques and more specifically, deep-learning algorithms have been used to facilitate their generations. Promising results were found thanks to Recurrent Neural Networks (RNN) trained on manually generated playlist and paired with clusters of extracted features from songs. We used the same principles to create the proposed algorithm but applied to a challenging medium, posters. First, a convolutional autoencoder was trained to extract features of the posters. The 52’000 digital posters were used as a training set. Poster features were then clustered. Next, an RNN learned to predict the next cluster according to the previous ones. RNN training set was composed of poster sequences extracted from a collection of books from the Gestaltung Museum of Zurich dedicated to displaying posters. Finally, within the predicted cluster, the poster with the best proximity compared to the previous poster is selected. The mean square distance between features of posters was used to compute the proximity. To validate the predictive model, we compared sequences of 15 posters produced by our model to randomly and manually generated sequences. Manual sequences were created by a professional graphic designer. We asked 21 participants working as professional graphic designers to sort the sequences from the one with the strongest graphic line to the one with the weakest and to motivate their answer with a short description. The sequences produced by the designer were ranked first 60%, second 25% and third 15% of the time. The sequences produced by our predictive model were ranked first 25%, second 45% and third 30% of the time. The sequences produced randomly were ranked first 15%, second 29%, and third 55% of the time. Compared to designer sequences, and as reported by participants, model and random sequences lacked thematic continuity. According to the results, the proposed model is able to generate better poster sequencing compared to random sampling. Eventually, our algorithm is sometimes able to outperform a professional designer. As a next step, the proposed algorithm should include a possibility to create sequences according to a selected theme. To conclude, this work shows the potentiality of artificial intelligence techniques to learn from existing content and provide a tool to curate large sets of data, with a permanent renewal of the presented content.

Keywords: Artificial Intelligence, Digital Humanities, serendipity, design research

Procedia PDF Downloads 186