Pub Date : 2019-12-01DOI: 10.1109/SSCI44817.2019.9002659
Abdulaziz T. Almutairi, J. Fieldsend
Recent work on multi-resolution optimisation (varying the fidelity of a design during a search) has developed approaches for automated resolution change depending on the population characteristics. This used the standard deviation of the population, or the marginal probability density estimation per variable, to automatically determine the resolution to apply to a design in the next generation. Here we build on this methodology in a number of new directions. We investigate the use of a complete estimated probability density function for resolution determination, enabling the dependencies between variables to be represented. We also explore the use of the multi-resolution transformation to assign a surrogate fitness to population members, but without modifying their location, and discuss the fitness landscape implications of this approach. Results are presented on a range of popular uni-objective continuous test-functions. These demonstrate the performance improvements that can be gained using an automated multi-resolution approach, and surprisingly indicate the simplest resolution indicator is often the most effective, but that relative performance is often problem dependant. We also observe how population duplicates grow in multi-resolution approaches, and discuss the implications of this when comparing algorithms (and efficiently implementing them).
{"title":"Automated and Surrogate Multi-Resolution Approaches in Genetic Algorithms","authors":"Abdulaziz T. Almutairi, J. Fieldsend","doi":"10.1109/SSCI44817.2019.9002659","DOIUrl":"https://doi.org/10.1109/SSCI44817.2019.9002659","url":null,"abstract":"Recent work on multi-resolution optimisation (varying the fidelity of a design during a search) has developed approaches for automated resolution change depending on the population characteristics. This used the standard deviation of the population, or the marginal probability density estimation per variable, to automatically determine the resolution to apply to a design in the next generation. Here we build on this methodology in a number of new directions. We investigate the use of a complete estimated probability density function for resolution determination, enabling the dependencies between variables to be represented. We also explore the use of the multi-resolution transformation to assign a surrogate fitness to population members, but without modifying their location, and discuss the fitness landscape implications of this approach. Results are presented on a range of popular uni-objective continuous test-functions. These demonstrate the performance improvements that can be gained using an automated multi-resolution approach, and surprisingly indicate the simplest resolution indicator is often the most effective, but that relative performance is often problem dependant. We also observe how population duplicates grow in multi-resolution approaches, and discuss the implications of this when comparing algorithms (and efficiently implementing them).","PeriodicalId":6729,"journal":{"name":"2019 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"3 1","pages":"2066-2073"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75410342","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-12-01DOI: 10.1109/SSCI44817.2019.9002991
Yong Fu, Lan Huang, Sujie Li, F. Lee
An innovative concept of mixing soil around obstacles via a newly developed equipment with articulated joints is currently being developed by a research group in NUS. To assist and supervise the movement of articulated joints in the real site, a supervisory software for trajectory mapping for the hardware of articulated joints is required. This study compares four approaches of controlling and optimizing for the subterranean trajectory of a three-jointed articulated deep mixing equipment, based on a proposed mathematical model, which is used to trace coordinates of joints in each step. Through the understanding of the equipment’s behaviour, the programs of deep soil mixing trajectory calculation using different methodologies, are developed and improved to achieve more functions and better performance. Comparison and analysis of all the feasible methodologies are included to obtain the optimal program for real world applications.
{"title":"Some Possible Trajectory Planning Methodologies for an Articulated Deep Soil Mixing Equipment with Limited Degrees-of-Freedom","authors":"Yong Fu, Lan Huang, Sujie Li, F. Lee","doi":"10.1109/SSCI44817.2019.9002991","DOIUrl":"https://doi.org/10.1109/SSCI44817.2019.9002991","url":null,"abstract":"An innovative concept of mixing soil around obstacles via a newly developed equipment with articulated joints is currently being developed by a research group in NUS. To assist and supervise the movement of articulated joints in the real site, a supervisory software for trajectory mapping for the hardware of articulated joints is required. This study compares four approaches of controlling and optimizing for the subterranean trajectory of a three-jointed articulated deep mixing equipment, based on a proposed mathematical model, which is used to trace coordinates of joints in each step. Through the understanding of the equipment’s behaviour, the programs of deep soil mixing trajectory calculation using different methodologies, are developed and improved to achieve more functions and better performance. Comparison and analysis of all the feasible methodologies are included to obtain the optimal program for real world applications.","PeriodicalId":6729,"journal":{"name":"2019 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"21 1","pages":"886-892"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78019858","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-12-01DOI: 10.1109/SSCI44817.2019.9002741
Gustavo B. M. Mello, S. Pontes-Filho, I. Sandvig, V. Valderhaug, E. Zouganeli, Ola Huse Ramstad, A. Sandvig, S. Nichele
In the brain, the structure of a network of neurons defines how these neurons implement the computations that underlie the mind and the behavior of animals and humans. Provided that we can describe the network of neurons as a graph. We can employ methods from graph theory to investigate its structure or use cellular automata to mathematically assess its function. Additionally, these graphs can provide biologically plausible designs for networks, which can be integrated as reservoirs to support computing. Although, software for the analysis of graphs and cellular automata are widely available. Graph extraction from the image of networks of brain cells remains difficult. Nervous tissue is heterogeneous, and differences in anatomy may reflect relevant differences in function. Here we introduce a deep learning based toolbox to extracts graphs from images of brain tissue. This toolbox provides an easy- to-use framework allowing system neuroscientists to generate graphs based on images of brain tissue by combining methods from image processing, deep learning, and graph theory. The goals are to simplify the training and usage of deep learning methods for computer vision and facilitate its integration into graph extraction pipelines. In this way, the toolbox provides an alternative to the required laborious manual process of tracing, sorting and classifying. We expect to democratize the machine learning methods to a wider community of users beyond the computer vision experts and improve the time-efficiency of graph extraction from large brain image datasets, which may lead to further understanding of the human mind.
{"title":"Method to Obtain Neuromorphic Reservoir Networks from Images of in Vitro Cortical Networks","authors":"Gustavo B. M. Mello, S. Pontes-Filho, I. Sandvig, V. Valderhaug, E. Zouganeli, Ola Huse Ramstad, A. Sandvig, S. Nichele","doi":"10.1109/SSCI44817.2019.9002741","DOIUrl":"https://doi.org/10.1109/SSCI44817.2019.9002741","url":null,"abstract":"In the brain, the structure of a network of neurons defines how these neurons implement the computations that underlie the mind and the behavior of animals and humans. Provided that we can describe the network of neurons as a graph. We can employ methods from graph theory to investigate its structure or use cellular automata to mathematically assess its function. Additionally, these graphs can provide biologically plausible designs for networks, which can be integrated as reservoirs to support computing. Although, software for the analysis of graphs and cellular automata are widely available. Graph extraction from the image of networks of brain cells remains difficult. Nervous tissue is heterogeneous, and differences in anatomy may reflect relevant differences in function. Here we introduce a deep learning based toolbox to extracts graphs from images of brain tissue. This toolbox provides an easy- to-use framework allowing system neuroscientists to generate graphs based on images of brain tissue by combining methods from image processing, deep learning, and graph theory. The goals are to simplify the training and usage of deep learning methods for computer vision and facilitate its integration into graph extraction pipelines. In this way, the toolbox provides an alternative to the required laborious manual process of tracing, sorting and classifying. We expect to democratize the machine learning methods to a wider community of users beyond the computer vision experts and improve the time-efficiency of graph extraction from large brain image datasets, which may lead to further understanding of the human mind.","PeriodicalId":6729,"journal":{"name":"2019 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"155 1","pages":"2360-2366"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74820655","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-12-01DOI: 10.1109/SSCI44817.2019.9003058
Kommalapati Sahil, A. K. Bhattacharya
Complex governing equations of physical phenomena like the Navier-Stokes' or Maxwell's equations can be numerically solved to yield detailed information on the characteristic variables of a process in the process domain interior, when the values at the boundary are known. This cannot be achieved in real time making it unamenable to achieve true benefits under Industry 4.0 where measured variables are available instantaneously at process boundaries but information in the domain interior is unobtainable for monitoring, control and optimization functions. The Universal Approximation Theorem provides a unique capability to Artificial Neural Networks - the ability to replicate the functionality of arbitrarily complex functions - including those represented by the above governing equations. A trained ANN can in principle replicate this functionality with high accuracy in milliseconds - hence can serve as the method of choice in Industry 4.0 frameworks to acquire characteristic process variables within the domain interior when boundary values are known from sensory inputs. This is however a concept still to be proven. This work intends to demonstrate this principle through numerical experimentation on a physical example that can be easily generalized.
{"title":"Accurate Replication of Simulations of Governing Equations of Processes in Industry 4.0 Environments with ANNs for Enhanced Monitoring and Control","authors":"Kommalapati Sahil, A. K. Bhattacharya","doi":"10.1109/SSCI44817.2019.9003058","DOIUrl":"https://doi.org/10.1109/SSCI44817.2019.9003058","url":null,"abstract":"Complex governing equations of physical phenomena like the Navier-Stokes' or Maxwell's equations can be numerically solved to yield detailed information on the characteristic variables of a process in the process domain interior, when the values at the boundary are known. This cannot be achieved in real time making it unamenable to achieve true benefits under Industry 4.0 where measured variables are available instantaneously at process boundaries but information in the domain interior is unobtainable for monitoring, control and optimization functions. The Universal Approximation Theorem provides a unique capability to Artificial Neural Networks - the ability to replicate the functionality of arbitrarily complex functions - including those represented by the above governing equations. A trained ANN can in principle replicate this functionality with high accuracy in milliseconds - hence can serve as the method of choice in Industry 4.0 frameworks to acquire characteristic process variables within the domain interior when boundary values are known from sensory inputs. This is however a concept still to be proven. This work intends to demonstrate this principle through numerical experimentation on a physical example that can be easily generalized.","PeriodicalId":6729,"journal":{"name":"2019 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"15 1","pages":"1873-1880"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74272905","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-12-01DOI: 10.1109/SSCI44817.2019.9002834
James Gillespie, I. Rañó, N. Siddique, Jose A. Santos, M. Khamassi
Braitenberg vehicles are bio-inspired controllers for sensor-based local navigation of wheeled robots that have been used in multiple real world robotic implementations. The common approach to implement such non-linear control mechanisms is through neural networks connecting sensing to motor action, yet tuning the weights to obtain appropriate closed-loop navigation behaviours can be very challenging. Standard approaches used hand tuned spiking or recurrent neural networks, or learnt the weights of feedforward networks using evolutionary approaches. Recently, Reinforcement Learning has been used to learn neural controllers for simulated Braitenberg vehicle 3a – a bio-inspired model of target seeking for wheeled robots – under the assumption of noiseless sensors. Real sensors, however, are subject to different levels of noise, and multiple works have shown that Braitenberg vehicles work even on outdoor robots, demonstrating that these control mechanisms work in harsh and dynamic environments. This paper shows that a robust neural controller for Braitenberg vehicle 3a can be learnt using policy gradient reinforcement learning in scenarios where sensor noise plays a non negligible role. The learnt controller is robust and tries to attenuate the effects of noise in the closed- loop navigation behaviour of the simulated stochastic vehicle. We compare the neural controller learnt using Reinforcement Learning with a simple hand tuned controller and show how the neural control mechanism outperforms a naïve controller. Results are illustrated through computer simulations of the closed-loop stochastic system.
{"title":"Using Reinforcement Learning to Attenuate for Stochasticity in Robot Navigation Controllers","authors":"James Gillespie, I. Rañó, N. Siddique, Jose A. Santos, M. Khamassi","doi":"10.1109/SSCI44817.2019.9002834","DOIUrl":"https://doi.org/10.1109/SSCI44817.2019.9002834","url":null,"abstract":"Braitenberg vehicles are bio-inspired controllers for sensor-based local navigation of wheeled robots that have been used in multiple real world robotic implementations. The common approach to implement such non-linear control mechanisms is through neural networks connecting sensing to motor action, yet tuning the weights to obtain appropriate closed-loop navigation behaviours can be very challenging. Standard approaches used hand tuned spiking or recurrent neural networks, or learnt the weights of feedforward networks using evolutionary approaches. Recently, Reinforcement Learning has been used to learn neural controllers for simulated Braitenberg vehicle 3a – a bio-inspired model of target seeking for wheeled robots – under the assumption of noiseless sensors. Real sensors, however, are subject to different levels of noise, and multiple works have shown that Braitenberg vehicles work even on outdoor robots, demonstrating that these control mechanisms work in harsh and dynamic environments. This paper shows that a robust neural controller for Braitenberg vehicle 3a can be learnt using policy gradient reinforcement learning in scenarios where sensor noise plays a non negligible role. The learnt controller is robust and tries to attenuate the effects of noise in the closed- loop navigation behaviour of the simulated stochastic vehicle. We compare the neural controller learnt using Reinforcement Learning with a simple hand tuned controller and show how the neural control mechanism outperforms a naïve controller. Results are illustrated through computer simulations of the closed-loop stochastic system.","PeriodicalId":6729,"journal":{"name":"2019 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"19 1","pages":"705-713"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74308539","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-12-01DOI: 10.1109/SSCI44817.2019.9003164
Huifang Yao, Hong He, Shilong Wang, Z. Xie
With the fast development of human-machine interface technology, emotion recognition has attracted more and more attentions in recent years. Compared to other physiological experimental signals frequently used in emotion recognition, EEG signals are easy to record but not easy to disguise. However, because of high dimensionality of EEG data and the diversity of human emotions, feature extraction and classification of EEG signals are still difficult. In this paper, we propose deep forest with multi-scale window (MSWDF) to identify EEG emotions. Deep Forest is an integrated method of decision trees. In the MSWDF, features can be extracted by multi-granularity scanning with multi-scale windows. Compared with deep neural network, the MSWDF not only has less parameters to adjust, but also can realize the classification of the dataset with small samples. In the MSWDF, raw EEG signals were firstly filtered and segmented into samples. Regarding EEG signals as multivariate time series, a new multi-granularity scanning strategy with variable windows is proposed to extract features from EEG samples. After classifying EEG features by the cascade forest, the recognition results are compared with these of Nearest Neighbor algorithm (KNN), Naive Bayes, Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVM). We found that the average classification accuracy of three emotions reaches to 84.90%, which is better than those of five compared methods.
{"title":"EEG-based Emotion Recognition Using Multi-scale Window Deep Forest","authors":"Huifang Yao, Hong He, Shilong Wang, Z. Xie","doi":"10.1109/SSCI44817.2019.9003164","DOIUrl":"https://doi.org/10.1109/SSCI44817.2019.9003164","url":null,"abstract":"With the fast development of human-machine interface technology, emotion recognition has attracted more and more attentions in recent years. Compared to other physiological experimental signals frequently used in emotion recognition, EEG signals are easy to record but not easy to disguise. However, because of high dimensionality of EEG data and the diversity of human emotions, feature extraction and classification of EEG signals are still difficult. In this paper, we propose deep forest with multi-scale window (MSWDF) to identify EEG emotions. Deep Forest is an integrated method of decision trees. In the MSWDF, features can be extracted by multi-granularity scanning with multi-scale windows. Compared with deep neural network, the MSWDF not only has less parameters to adjust, but also can realize the classification of the dataset with small samples. In the MSWDF, raw EEG signals were firstly filtered and segmented into samples. Regarding EEG signals as multivariate time series, a new multi-granularity scanning strategy with variable windows is proposed to extract features from EEG samples. After classifying EEG features by the cascade forest, the recognition results are compared with these of Nearest Neighbor algorithm (KNN), Naive Bayes, Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVM). We found that the average classification accuracy of three emotions reaches to 84.90%, which is better than those of five compared methods.","PeriodicalId":6729,"journal":{"name":"2019 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"24 1","pages":"381-386"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"72621998","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-12-01DOI: 10.1109/SSCI44817.2019.9003013
Xinxing Duan, Sujia Yin
Ultrasound-based drug delivery inspires rigorous experimental studies on the cell experiments involving microbubble-mediated ultrasound. Yet few tools can be found to accelerate the cellular exposure experiments on the heterogeneous-population-based bioeffects under various conditions. In this study, we propose a 2D model which can be used to simulate the population evolution of the co-cultured cells after microbubble-mediated ultrasound is induced. Two update rules based on the experimental data were proposed and incorporated in the simulation procedures. The bioeffects found in the previous experiment, such as cell cycle arrest, are taken into account in the model to simulate the population growth in a co-cultured setting. This model may provide a predictive tool for the multi-cell type responses to ultrasound-induced perforation and facilitate the future experiment design.
{"title":"Modeling the Population Growth of the Co-cultured Blood Cells Exposed by Microbubble-mediated Ultrasound","authors":"Xinxing Duan, Sujia Yin","doi":"10.1109/SSCI44817.2019.9003013","DOIUrl":"https://doi.org/10.1109/SSCI44817.2019.9003013","url":null,"abstract":"Ultrasound-based drug delivery inspires rigorous experimental studies on the cell experiments involving microbubble-mediated ultrasound. Yet few tools can be found to accelerate the cellular exposure experiments on the heterogeneous-population-based bioeffects under various conditions. In this study, we propose a 2D model which can be used to simulate the population evolution of the co-cultured cells after microbubble-mediated ultrasound is induced. Two update rules based on the experimental data were proposed and incorporated in the simulation procedures. The bioeffects found in the previous experiment, such as cell cycle arrest, are taken into account in the model to simulate the population growth in a co-cultured setting. This model may provide a predictive tool for the multi-cell type responses to ultrasound-induced perforation and facilitate the future experiment design.","PeriodicalId":6729,"journal":{"name":"2019 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"22 1","pages":"293-298"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84941578","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-12-01DOI: 10.1109/SSCI44817.2019.9002940
Julie L. Harvey, S. Kumar
Data science is a field that can be used in a variety of settings. Education is one of the fields that is expanding its use of data science to improve the quality of education. The United States denotes primary and secondary school as grades kindergarten (K) through 12th grade. This is representative of education prior to college/university level. Data science in K-12 education is evaluated and important to the field of education because educators, administrators, and stakeholders are always looking for ways to close the achievement gap and increase performance of all students. Student performance evaluation using data science is crucial to closing this gap. Data mining is used in the evaluation and analysis of student performance, educational programs and educational instruction. It is also used to create prediction models for future student success. A K-12 education dataset will be used to evaluate student performance. This paper will explore and display student performance based on a variety of factors and data. Data science in K-12 education and its impact on student performance and educator use of this data is discussed. We have also performed review of existing work in the data analytics for K-12 education along with their limitations.
{"title":"Data Science for K-12 Education","authors":"Julie L. Harvey, S. Kumar","doi":"10.1109/SSCI44817.2019.9002940","DOIUrl":"https://doi.org/10.1109/SSCI44817.2019.9002940","url":null,"abstract":"Data science is a field that can be used in a variety of settings. Education is one of the fields that is expanding its use of data science to improve the quality of education. The United States denotes primary and secondary school as grades kindergarten (K) through 12th grade. This is representative of education prior to college/university level. Data science in K-12 education is evaluated and important to the field of education because educators, administrators, and stakeholders are always looking for ways to close the achievement gap and increase performance of all students. Student performance evaluation using data science is crucial to closing this gap. Data mining is used in the evaluation and analysis of student performance, educational programs and educational instruction. It is also used to create prediction models for future student success. A K-12 education dataset will be used to evaluate student performance. This paper will explore and display student performance based on a variety of factors and data. Data science in K-12 education and its impact on student performance and educator use of this data is discussed. We have also performed review of existing work in the data analytics for K-12 education along with their limitations.","PeriodicalId":6729,"journal":{"name":"2019 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"7 1","pages":"2482-2488"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81981861","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-12-01DOI: 10.1109/SSCI44817.2019.9002753
Yifan Xu, Dongrui Wu
Drowsy driving is one of the major contributors to traffic accidents. Continuously detecting the driver’s drowsiness and taking actions accordingly may be one solution to improving driving safety. Electroencephalogram (EEG) signals contain information of the brain state, and hence can be utilized to estimate the driver’s drowsiness level. A challenge in EEG-based drowsiness estimation is that when directly applied to a new subject without any calibration, the system’s performance usually degrades significantly. Many efforts have been devoted to reducing the calibration data requirement, but there are still very few approaches that can completely eliminate the calibration process. This paper proposes a self-paced learning approach, which also takes the label diversity into consideration. The model learns from the easiest samples when the training first starts, and then more difficult ones are gradually added to the training process. This training strategy improves the generalization performance of the model on a new subject. Experiments on a simulated driving dataset with 15 subjects demonstrated that the proposed approach can better reduce the estimation error than several other approaches.
{"title":"EEG-Based Driver Drowsiness Estimation Using Self-Paced Learning with Label Diversity","authors":"Yifan Xu, Dongrui Wu","doi":"10.1109/SSCI44817.2019.9002753","DOIUrl":"https://doi.org/10.1109/SSCI44817.2019.9002753","url":null,"abstract":"Drowsy driving is one of the major contributors to traffic accidents. Continuously detecting the driver’s drowsiness and taking actions accordingly may be one solution to improving driving safety. Electroencephalogram (EEG) signals contain information of the brain state, and hence can be utilized to estimate the driver’s drowsiness level. A challenge in EEG-based drowsiness estimation is that when directly applied to a new subject without any calibration, the system’s performance usually degrades significantly. Many efforts have been devoted to reducing the calibration data requirement, but there are still very few approaches that can completely eliminate the calibration process. This paper proposes a self-paced learning approach, which also takes the label diversity into consideration. The model learns from the easiest samples when the training first starts, and then more difficult ones are gradually added to the training process. This training strategy improves the generalization performance of the model on a new subject. Experiments on a simulated driving dataset with 15 subjects demonstrated that the proposed approach can better reduce the estimation error than several other approaches.","PeriodicalId":6729,"journal":{"name":"2019 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"46 1","pages":"369-375"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82086957","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-12-01DOI: 10.1109/SSCI44817.2019.9002669
Vatsalya Chaubey, Maya S. Nair, G. Pillai
Histones are proteins around which DNA is coiled to form chromatin fibres in the nucleus of a biological cell. They undergo modifications post-translation, a process which produces proteins using the DNA as the blueprint. These modifications play a very important role in regulating gene expression by influencing the translation process. The knowledge of how such modifications affect gene expression and the need for an accurate pipeline to predict the expression values from modification signals is undeniable. In this paper, we present the first generalized deep learning model to classify gene expression based on the histone modification signals irrespective of the type of cell from which the signal was recorded. Our approach automatically performs feature extraction using ID convolutional layers which are used further to establish relationships among the learned features and make accurate predictions. This model is able to make predictions on all the different cell types by being trained only once. It also outperforms the present state of the art when compared against the predictions made for different kinds of cells and the computational resources required.
{"title":"Gene Expression Prediction Using a Deep 1D Convolution Neural Network","authors":"Vatsalya Chaubey, Maya S. Nair, G. Pillai","doi":"10.1109/SSCI44817.2019.9002669","DOIUrl":"https://doi.org/10.1109/SSCI44817.2019.9002669","url":null,"abstract":"Histones are proteins around which DNA is coiled to form chromatin fibres in the nucleus of a biological cell. They undergo modifications post-translation, a process which produces proteins using the DNA as the blueprint. These modifications play a very important role in regulating gene expression by influencing the translation process. The knowledge of how such modifications affect gene expression and the need for an accurate pipeline to predict the expression values from modification signals is undeniable. In this paper, we present the first generalized deep learning model to classify gene expression based on the histone modification signals irrespective of the type of cell from which the signal was recorded. Our approach automatically performs feature extraction using ID convolutional layers which are used further to establish relationships among the learned features and make accurate predictions. This model is able to make predictions on all the different cell types by being trained only once. It also outperforms the present state of the art when compared against the predictions made for different kinds of cells and the computational resources required.","PeriodicalId":6729,"journal":{"name":"2019 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":"119 1","pages":"1383-1389"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79409909","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：