Background: Classification using microarray datasets is usually based on a small number of sample... more Background: Classification using microarray datasets is usually based on a small number of samples for which tens of thousands of gene expression measurements have been obtained. The selection of the genes most significant to the classification problem is a challenging issue in high dimension data analysis and interpretation. A previous study with SVM-RCE (Recursive Cluster Elimination), suggested that classification based on groups of correlated genes sometimes exhibits better performance than classification using single genes. Large databases of gene interaction networks provide an important resource for the analysis of genetic phenomena and for classification studies using interacting genes. We now demonstrate that an algorithm which integrates network information with recursive feature elimination based on SVM exhibits good performance and improves the biological interpretability of the results. We refer to the method as SVM with Recursive Network Elimination (SVM-RNE) Results: Initially, one thousand genes selected by t-test from a training set are filtered so that only genes that map to a gene network database remain. The Gene Expression Network Analysis Tool (GXNA) is applied to the remaining genes to form n clusters of genes that are highly connected in the network. Linear SVM is used to classify the samples using these clusters, and a weight is assigned to each cluster based on its importance to the classification. The least informative clusters are removed while retaining the remainder for the next classification step. This process is repeated until an optimal classification is obtained. Conclusion: More than 90% accuracy can be obtained in classification of selected microarray datasets by integrating the interaction network information with the gene expression information from the microarrays.
In recent work, attacks on automated classification of Parkinson’s disease have encountered diffi... more In recent work, attacks on automated classification of Parkinson’s disease have encountered difficulties, especially for cross-individual generalization. This is crucial since (i) Classifying the degree of Parkinson’s disease is an important clinical necessity. (ii) The lack of such an automated system leaves current clinical methodology to use manual and subjective classification by a trained clinician. In earlier work, two of the authors of this paper have shown that, directly from the speech signal, reliable classification as to the presence of the disease can be produced using a machine learning approach. However, this approach was unable to reliably classify the severity degree of the disease. In other work, a deep (convolutional) neural network was tried on the same data set (albeit without feature extraction), which again did not succeed on the multi-label case.In this work, we applied a data science approach to solve this problem by analysing the topological structure of the label space and the internal topological structure of the data. Specifically we explored using (i) the linearity of the label-space to reduce the inherent noise in multi-class classifiers and (ii) to break the data into separate topological clusters (by using a version of unsupervised topological learning) and then applying separate classification parametrizations for each cluster.While our interest was mainly directed to the Parkinson’s classification problem, the methods seem relatively generic and should be applicable to many data sets. (As an example, we also applied this directly to a well-known baseline data set - wine classification and obtained state of the art results).On the Parkinson classification task, these methods obtained, on a 7 degree classification scale, results which are comparable to the best accuracy on simple two class classification.
Annals of Mathematics and Artificial Intelligence, Aug 28, 2019
In the past few years, due to their ability to extract multivariate correlations, machine learnin... more In the past few years, due to their ability to extract multivariate correlations, machine learning tools have become more and more important for discovery of information in very complex data sets. This has had specific application to various data sets related to human brain tasks. However, this is far from a simple and direct methodology. Some of the issues involve dealing with the extreme signal to noise ratios, as well as variation between different individuals. Moreover, the huge amount of features relative to the number of data points is a challenge. As a result, in attacking these problems, we found it necessary to adapt a large variety of methodologies; chosen to overcome specific obstructions for specific problems. In this paper, we describe our experience working on several examples at the edge of capabilities of these systems and describe the various and variant methodologies we needed to overcome these sort of challenges. Hopefully these cases will serve as a guideline for other applications.
A novel fMRI classification method designed for rapid event related fMRI experiments is described... more A novel fMRI classification method designed for rapid event related fMRI experiments is described and applied to the classification of loud reading of isolated words in Hebrew. Three comparisons of different grammatical complexity were performed: (i) words versus asterisks (ii) “with diacritics versus without diacritics” and (iii) “with root versus no root”. We discuss the most difficult task and, for comparison, the easiest one. Earlier work using more standard classification techniques (machine learning and statistical) succeeded fully only in the simplest of these tasks (i), but produced only partial results on (ii) and failed completely, even on the training set on the deepest task (iii). The method performs a “best match” between pre-processed data and computing a full library of artificially generated examples. The method involves a deconvolution of the rapid events on the data and performing a convolution on the generated data. The best-match is performed over all “words” constructed by convolving the response functions of each value of each event performed in a “windowed” sequence. This is accomplished separately for all voxels and then a voting procedure defines the outcome. Using the same feature selection (ANOVA) as in the earlier methods, (i) there is a dramatic increase in the accuracy rate for the third (most difficult task) on the intra-run level (88%) as well as the first task (ii) Unlike the earlier methods training and testing over all runs (within subject) achieves a significant level of classification (64% accuracy) for the training set. This shows the information for this “deeper” cognitive task can in fact be extracted from the fMRI information.
Classifying the degree of Parkinson’s disease is an important clinical necessity. Nonetheless, cu... more Classifying the degree of Parkinson’s disease is an important clinical necessity. Nonetheless, current methodology requires manual (and subjective) evaluation by a trained clinical expert. Recently, Machine Learning tools have been developed that can produce a classification of the presence of PD directly from the speech signal in an automated and objective fashion. However, these methods were not sufficient for the classification of the degree of the disease. In this work, we show how to apply and leverage topological information on the both the label space and the feature space of the speech signal in order to solve this problem.We address the problem by performing topological clustering (using a version of the Kohonen Self Organizing Map algorithm) of the feature space and then optimizing separate multi-class classifiers on each cluster.Using these methods, we can reliably train our system to classify new speech signal data to more than the 70% level on a 7 degree classification (where random level is 14%) which is close to the obtainable accuracy on the simple 2 class classification.
International Journal of Pattern Recognition and Artificial Intelligence, Mar 1, 2007
Collaborative Filtering (CF) is currently one of the most popular and most widely used personaliz... more Collaborative Filtering (CF) is currently one of the most popular and most widely used personalization techniques. It generates personalized predictions based on the assumption that users with similar tastes prefer similar items. One of the major drawbacks of the CF from the computational point of view is its limited scalability since the computational effort required by the CF grows linearly both with the number of available users and items. This work proposes a novel efficient variant of the CF employed over a multidimensional content-addressable space. The proposed approach heuristically decreases the computational effort required by the CF algorithm by limiting the search process only to potentially similar users. Experimental results demonstrate that the proposed heuristic approach is capable of generating predictions with high levels of accuracy, while significantly improving the performance in comparison with the traditional implementations of the CF.
2016 IEEE International Conference on the Science of Electrical Engineering (ICSEE), 2016
Parkinson's Disease (PD) is a relatively common neurodegenerative disabling disease. It affec... more Parkinson's Disease (PD) is a relatively common neurodegenerative disabling disease. It affects central nervous system with profound effect on the motor system. The most common symptoms include slowness, rigidity and tremor during motion. It has been suggested that the vocal cords are among the first one to be affected and thus the speech is affected at very early stage of the disease and continues to deteriorate as the disease progress. In this work, we focus on automating the process of diagnosis from continuous native speech by removing the necessity of a trained personal from the diagnosis process. This is done by using an adaptation of Convolutional Neural Network (CNN) architecture for one-dimensional signal processing (i.e. raw speech signal) on a relatively small training set. This is a continuation to previous works where we showed (i) that this task can be achieved by using manually-extracted features of the speech (such as formants and their ratios) and (ii) by using an automatic process of auditory features extraction, where the features were selected by signal processing specialist.
Transactions on Computational Collective Intelligence XXIV, 2016
This work use supervised machine learning methods on fMRI brain scans, taken/measured during a me... more This work use supervised machine learning methods on fMRI brain scans, taken/measured during a memory-retrieval task, to support establishing the existence of two distinct systems for human declarative memory "Explicit Encoding" EE and "Fast Mapping" FM. The importance of using retrieval is that it allows a direct comparison between exemplars designed to use EE and those designed to use FM. This is not directly available under acquisition tasks because of the nature of the purported memory systems since the tasks are necessarily somewhat distinct between the two systems under acquisition. This means that there could be a confounding of the distinction in the task with the difference in the representation and mechanism of the internal memory system during analysis. Retrieval tasks, on the other hand allow for identity of task. Thus this work fills a lacuna in earlier work which used memory acquisition tasks. In addition, since the data used in this work was gathered over a two day period, the classification methods is also able to identify a distinction in the consolidation of the memories in the two systems. The results presented here clearly support the existence of the two distinct memory systems.
Proceedings of the 7th International Joint Conference on Computational Intelligence, 2015
This work uses supervised machine learning methods over fMRI brain scans to establish the existen... more This work uses supervised machine learning methods over fMRI brain scans to establish the existence of two different encoding procedures for human declarative memory. Declarative knowledge refers to the memory for facts and events and initially depends on the hippocampus. Recent studies which used patients with hippocampal lesions and neuroimaging data, suggested the existence of an alternative process to form declarative memories. This process is triggered by learning mechanism called "Fast Mapping (FM)", as opposed to the 'standard' "Explicit Encoding (EE)" learning procedure. The present work gives a clear biomarker on the existence of two distinct encoding procedures as we can accurately predict which of the processes is being used directly from voxel activity in fMRI scans. The scans are taken during retrieval of information wherein the tasks are identical regardless of which procedure was used for acquisition and by that reflect conclusive prediction. This is an identification of a more subtle cognitive task than direct perceptual cognitive tasks as it requires some encoding and processing in the brain.
We show that fMRI analysis using machine learning tools are sufficient to distinguish valence (i.... more We show that fMRI analysis using machine learning tools are sufficient to distinguish valence (i.e., positive or negative) of freely retrieved autobiographical memories in a cross-participant setting. Our methodology uses feature selection (ReliefF) in combination with boosting methods, both applied directly to data represented in voxel space. In previous work using the same data set, Nawa and Ando showed that whole-brain based classification could achieve above-chance classification accuracy only when both training and testing data came from the same individual. In a cross-participant setting, classification results were not statistically significant. Additionally, on average the classification accuracy obtained when using ReliefF is substantially higher than previous results-81% for the within-participant classification, and 62% for the cross-participant classification. Furthermore, since features are defined in voxel space, it is possible to show brain maps indicating the regions of that are most relevant in determining the results of the classification. Interestingly, the voxels that were selected using the proposed computational pipeline seem to be consistent with current neurophysiological theories regarding the brain regions actively involved in autobiographical memory processes.
In this talk, we will illustrate with two examples how temporal activity naturally emerges from t... more In this talk, we will illustrate with two examples how temporal activity naturally emerges from the structure of what is a priori atemporal models. The two examples are (i) the process of rate selection in a history-dependent manner in a neuron and (ii) the occurrence of temporal associative memory, where the time falls out naturally as a result of noise in the system. The rst example presents the elements of a mathematical computational model that re ects the experimental nding that the time scale of a neuron is not xed; but rather varies with the history of its stimulus. Unlike most physiological models, there is no predetermined rates associated with transitions between states of the system nor are there predetermined constants associated with adaptation rates; instead the model is a kind of \modulating automata" where the rates emerge from the history of the system itself. We focus here on the temporal dynamics of a neuron and show how a simple internal structure will give ...
Accurate initialization of a User Model (UM) is im portant for every system that provides persona... more Accurate initialization of a User Model (UM) is im portant for every system that provides personalized services. Howeve r, th re are systems where this initialization is critical yet no user data is available from prior interactions. For example, one in which both the repeat usage by a user is rare, and the total interaction on a single use is relatively limited. Evaluating the effectiveness of a User Model and particular instance of such a model is never an easy task, particularly through ser studies. Moreover, evaluation typically focuses on the usability of an entire system rather than the performance of a specific UM instantiation. In thi s paper we propose evaluating the quality of UMs via simulation and co mparison to a "gold standard". This standard is an approximation of th e user's ideal model. We will demonstrate this through a case study of a museum's visitor guide system implemented in the Hecht Museum at the University o f Haifa, Israel
We show how one-class compression Neural Networks and one-class SVM can be applied to fMRI data t... more We show how one-class compression Neural Networks and one-class SVM can be applied to fMRI data to learn the classification of brain activity associated with a specific motor activity. For comparison purposes, we use two labeled data and see what degree of classification ability is lost compared with the usual two-class SVM.
2020 International Joint Conference on Neural Networks (IJCNN), 2020
We develop a method that is based on processing gathered Event Related Potentials (ERP) signals a... more We develop a method that is based on processing gathered Event Related Potentials (ERP) signals and the use of machine learning technique for multivariate analysis (i.e. classification) that we apply in order to analyze the differences between Dyslexic and Skilled readers.No human intervention is needed in the analysis process. This is the state of the art results for automatic identification of Dyslexic readers using a Lexical Decision Task. We use mathematical and machine learning based techniques to automatically discover novel complex features that (i) allow for reliable distinction between Dyslexic and Normal Control Skilled readers and (ii) to validate the assumption that most of the differences between Dyslexic and Skilled readers are located in the left hemisphere.Interestingly, these tools also pointed to the fact that High Pass signals (typically considered as "noise" during ERP/EEG analyses) in fact contain significant relevant information.Finally, the proposed ...
We implemented versions of the SVM appropriate for one-class classification in the context of inf... more We implemented versions of the SVM appropriate for one-class classification in the context of information retrieval. The experiments were conducted on the standard Reuters data set. For the SVM implementation we used both a version of Schoelkopf et al. and a somewhat different version of one-class SVM based on identifying "outlier" data as representative of the second-class. We report on experiments with different kernels for both of these implementations and with different representations of the data, including binary vectors, tf-idf representation and a modification called "Hadamard" representation. Then we compared it with one-class versions of the algorithms prototype (Rocchio), nearest neighbor, naive Bayes, and finally a natural one-class neural network classification method based on "bottleneck" compression generated filters.The SVM approach as represented by Schoelkopf was superior to all the methods except the neural network one, where it was, ...
We develop a method that is based on processing gathered Event Related Potentials (ERP) signals a... more We develop a method that is based on processing gathered Event Related Potentials (ERP) signals and the use of machine learning technique for multivariate analysis (i.e. classification) that we apply in order to analyze the differences between Dyslexic and Skilled readers. No human intervention is needed in the analysis process. This is the state of the art results for automatic identification of Dyslexic readers using a Lexical Decision Task. We use mathematical and machine learning based techniques to automatically discover novel complex features that (i) allow for reliable distinction between Dyslexic and Normal Control Skilled readers and (ii) to validate the assumption that the most of the differences between Dyslexic and Skilled readers located in the left hemisphere. Interestingly, these tools also pointed to the fact that High Pass signals (typically considered as "noise" during ERP/EEG analyses) in fact contains significant relevant information. Finally, the propo...
Recent advances in the DNA microarray technology have provided the ability to examine and measure... more Recent advances in the DNA microarray technology have provided the ability to examine and measure the expression levels of thousands of genes simultaneously in an organism. In this technology each gene is recorded under different conditions or each gene is evaluated under a single environment but in different types of tissues. In the first case it is used in identification of functionally related genes where asthe second type of technology is helpful in classification of different types of tissues and identification of those genes whose expression levels are good diagnostic indicators. Different approaches have been applied to classify different datasets. However, the main challenges in this task is the availability of a smaller number of samples compared to huge number of genes and the noisy nature of biological data. This paper review on different techniques used to classify the genes and improved efficiency of biomarker identification due to these classifications.
Using Machine Learning Tools (Neural Networks and Support Vector Machines) we show how raw fMRI b... more Using Machine Learning Tools (Neural Networks and Support Vector Machines) we show how raw fMRI brain scan data can be correctly assigned to cognitive tasks. We describe experiments classifying visual and motor tasks using one-class and two-class labeling for training. No a priori knowledge (e.g. of anatomy or physiology) is needed for the system to work. These results further suggest that feature reduction techniques may allow for the automatic location of brain areas correlated with specific cognitive tasks possibly evn when the needed features are not localized.
Background: Classification using microarray datasets is usually based on a small number of sample... more Background: Classification using microarray datasets is usually based on a small number of samples for which tens of thousands of gene expression measurements have been obtained. The selection of the genes most significant to the classification problem is a challenging issue in high dimension data analysis and interpretation. A previous study with SVM-RCE (Recursive Cluster Elimination), suggested that classification based on groups of correlated genes sometimes exhibits better performance than classification using single genes. Large databases of gene interaction networks provide an important resource for the analysis of genetic phenomena and for classification studies using interacting genes. We now demonstrate that an algorithm which integrates network information with recursive feature elimination based on SVM exhibits good performance and improves the biological interpretability of the results. We refer to the method as SVM with Recursive Network Elimination (SVM-RNE) Results: Initially, one thousand genes selected by t-test from a training set are filtered so that only genes that map to a gene network database remain. The Gene Expression Network Analysis Tool (GXNA) is applied to the remaining genes to form n clusters of genes that are highly connected in the network. Linear SVM is used to classify the samples using these clusters, and a weight is assigned to each cluster based on its importance to the classification. The least informative clusters are removed while retaining the remainder for the next classification step. This process is repeated until an optimal classification is obtained. Conclusion: More than 90% accuracy can be obtained in classification of selected microarray datasets by integrating the interaction network information with the gene expression information from the microarrays.
In recent work, attacks on automated classification of Parkinson’s disease have encountered diffi... more In recent work, attacks on automated classification of Parkinson’s disease have encountered difficulties, especially for cross-individual generalization. This is crucial since (i) Classifying the degree of Parkinson’s disease is an important clinical necessity. (ii) The lack of such an automated system leaves current clinical methodology to use manual and subjective classification by a trained clinician. In earlier work, two of the authors of this paper have shown that, directly from the speech signal, reliable classification as to the presence of the disease can be produced using a machine learning approach. However, this approach was unable to reliably classify the severity degree of the disease. In other work, a deep (convolutional) neural network was tried on the same data set (albeit without feature extraction), which again did not succeed on the multi-label case.In this work, we applied a data science approach to solve this problem by analysing the topological structure of the label space and the internal topological structure of the data. Specifically we explored using (i) the linearity of the label-space to reduce the inherent noise in multi-class classifiers and (ii) to break the data into separate topological clusters (by using a version of unsupervised topological learning) and then applying separate classification parametrizations for each cluster.While our interest was mainly directed to the Parkinson’s classification problem, the methods seem relatively generic and should be applicable to many data sets. (As an example, we also applied this directly to a well-known baseline data set - wine classification and obtained state of the art results).On the Parkinson classification task, these methods obtained, on a 7 degree classification scale, results which are comparable to the best accuracy on simple two class classification.
Annals of Mathematics and Artificial Intelligence, Aug 28, 2019
In the past few years, due to their ability to extract multivariate correlations, machine learnin... more In the past few years, due to their ability to extract multivariate correlations, machine learning tools have become more and more important for discovery of information in very complex data sets. This has had specific application to various data sets related to human brain tasks. However, this is far from a simple and direct methodology. Some of the issues involve dealing with the extreme signal to noise ratios, as well as variation between different individuals. Moreover, the huge amount of features relative to the number of data points is a challenge. As a result, in attacking these problems, we found it necessary to adapt a large variety of methodologies; chosen to overcome specific obstructions for specific problems. In this paper, we describe our experience working on several examples at the edge of capabilities of these systems and describe the various and variant methodologies we needed to overcome these sort of challenges. Hopefully these cases will serve as a guideline for other applications.
A novel fMRI classification method designed for rapid event related fMRI experiments is described... more A novel fMRI classification method designed for rapid event related fMRI experiments is described and applied to the classification of loud reading of isolated words in Hebrew. Three comparisons of different grammatical complexity were performed: (i) words versus asterisks (ii) “with diacritics versus without diacritics” and (iii) “with root versus no root”. We discuss the most difficult task and, for comparison, the easiest one. Earlier work using more standard classification techniques (machine learning and statistical) succeeded fully only in the simplest of these tasks (i), but produced only partial results on (ii) and failed completely, even on the training set on the deepest task (iii). The method performs a “best match” between pre-processed data and computing a full library of artificially generated examples. The method involves a deconvolution of the rapid events on the data and performing a convolution on the generated data. The best-match is performed over all “words” constructed by convolving the response functions of each value of each event performed in a “windowed” sequence. This is accomplished separately for all voxels and then a voting procedure defines the outcome. Using the same feature selection (ANOVA) as in the earlier methods, (i) there is a dramatic increase in the accuracy rate for the third (most difficult task) on the intra-run level (88%) as well as the first task (ii) Unlike the earlier methods training and testing over all runs (within subject) achieves a significant level of classification (64% accuracy) for the training set. This shows the information for this “deeper” cognitive task can in fact be extracted from the fMRI information.
Classifying the degree of Parkinson’s disease is an important clinical necessity. Nonetheless, cu... more Classifying the degree of Parkinson’s disease is an important clinical necessity. Nonetheless, current methodology requires manual (and subjective) evaluation by a trained clinical expert. Recently, Machine Learning tools have been developed that can produce a classification of the presence of PD directly from the speech signal in an automated and objective fashion. However, these methods were not sufficient for the classification of the degree of the disease. In this work, we show how to apply and leverage topological information on the both the label space and the feature space of the speech signal in order to solve this problem.We address the problem by performing topological clustering (using a version of the Kohonen Self Organizing Map algorithm) of the feature space and then optimizing separate multi-class classifiers on each cluster.Using these methods, we can reliably train our system to classify new speech signal data to more than the 70% level on a 7 degree classification (where random level is 14%) which is close to the obtainable accuracy on the simple 2 class classification.
International Journal of Pattern Recognition and Artificial Intelligence, Mar 1, 2007
Collaborative Filtering (CF) is currently one of the most popular and most widely used personaliz... more Collaborative Filtering (CF) is currently one of the most popular and most widely used personalization techniques. It generates personalized predictions based on the assumption that users with similar tastes prefer similar items. One of the major drawbacks of the CF from the computational point of view is its limited scalability since the computational effort required by the CF grows linearly both with the number of available users and items. This work proposes a novel efficient variant of the CF employed over a multidimensional content-addressable space. The proposed approach heuristically decreases the computational effort required by the CF algorithm by limiting the search process only to potentially similar users. Experimental results demonstrate that the proposed heuristic approach is capable of generating predictions with high levels of accuracy, while significantly improving the performance in comparison with the traditional implementations of the CF.
2016 IEEE International Conference on the Science of Electrical Engineering (ICSEE), 2016
Parkinson's Disease (PD) is a relatively common neurodegenerative disabling disease. It affec... more Parkinson's Disease (PD) is a relatively common neurodegenerative disabling disease. It affects central nervous system with profound effect on the motor system. The most common symptoms include slowness, rigidity and tremor during motion. It has been suggested that the vocal cords are among the first one to be affected and thus the speech is affected at very early stage of the disease and continues to deteriorate as the disease progress. In this work, we focus on automating the process of diagnosis from continuous native speech by removing the necessity of a trained personal from the diagnosis process. This is done by using an adaptation of Convolutional Neural Network (CNN) architecture for one-dimensional signal processing (i.e. raw speech signal) on a relatively small training set. This is a continuation to previous works where we showed (i) that this task can be achieved by using manually-extracted features of the speech (such as formants and their ratios) and (ii) by using an automatic process of auditory features extraction, where the features were selected by signal processing specialist.
Transactions on Computational Collective Intelligence XXIV, 2016
This work use supervised machine learning methods on fMRI brain scans, taken/measured during a me... more This work use supervised machine learning methods on fMRI brain scans, taken/measured during a memory-retrieval task, to support establishing the existence of two distinct systems for human declarative memory "Explicit Encoding" EE and "Fast Mapping" FM. The importance of using retrieval is that it allows a direct comparison between exemplars designed to use EE and those designed to use FM. This is not directly available under acquisition tasks because of the nature of the purported memory systems since the tasks are necessarily somewhat distinct between the two systems under acquisition. This means that there could be a confounding of the distinction in the task with the difference in the representation and mechanism of the internal memory system during analysis. Retrieval tasks, on the other hand allow for identity of task. Thus this work fills a lacuna in earlier work which used memory acquisition tasks. In addition, since the data used in this work was gathered over a two day period, the classification methods is also able to identify a distinction in the consolidation of the memories in the two systems. The results presented here clearly support the existence of the two distinct memory systems.
Proceedings of the 7th International Joint Conference on Computational Intelligence, 2015
This work uses supervised machine learning methods over fMRI brain scans to establish the existen... more This work uses supervised machine learning methods over fMRI brain scans to establish the existence of two different encoding procedures for human declarative memory. Declarative knowledge refers to the memory for facts and events and initially depends on the hippocampus. Recent studies which used patients with hippocampal lesions and neuroimaging data, suggested the existence of an alternative process to form declarative memories. This process is triggered by learning mechanism called "Fast Mapping (FM)", as opposed to the 'standard' "Explicit Encoding (EE)" learning procedure. The present work gives a clear biomarker on the existence of two distinct encoding procedures as we can accurately predict which of the processes is being used directly from voxel activity in fMRI scans. The scans are taken during retrieval of information wherein the tasks are identical regardless of which procedure was used for acquisition and by that reflect conclusive prediction. This is an identification of a more subtle cognitive task than direct perceptual cognitive tasks as it requires some encoding and processing in the brain.
We show that fMRI analysis using machine learning tools are sufficient to distinguish valence (i.... more We show that fMRI analysis using machine learning tools are sufficient to distinguish valence (i.e., positive or negative) of freely retrieved autobiographical memories in a cross-participant setting. Our methodology uses feature selection (ReliefF) in combination with boosting methods, both applied directly to data represented in voxel space. In previous work using the same data set, Nawa and Ando showed that whole-brain based classification could achieve above-chance classification accuracy only when both training and testing data came from the same individual. In a cross-participant setting, classification results were not statistically significant. Additionally, on average the classification accuracy obtained when using ReliefF is substantially higher than previous results-81% for the within-participant classification, and 62% for the cross-participant classification. Furthermore, since features are defined in voxel space, it is possible to show brain maps indicating the regions of that are most relevant in determining the results of the classification. Interestingly, the voxels that were selected using the proposed computational pipeline seem to be consistent with current neurophysiological theories regarding the brain regions actively involved in autobiographical memory processes.
In this talk, we will illustrate with two examples how temporal activity naturally emerges from t... more In this talk, we will illustrate with two examples how temporal activity naturally emerges from the structure of what is a priori atemporal models. The two examples are (i) the process of rate selection in a history-dependent manner in a neuron and (ii) the occurrence of temporal associative memory, where the time falls out naturally as a result of noise in the system. The rst example presents the elements of a mathematical computational model that re ects the experimental nding that the time scale of a neuron is not xed; but rather varies with the history of its stimulus. Unlike most physiological models, there is no predetermined rates associated with transitions between states of the system nor are there predetermined constants associated with adaptation rates; instead the model is a kind of \modulating automata" where the rates emerge from the history of the system itself. We focus here on the temporal dynamics of a neuron and show how a simple internal structure will give ...
Accurate initialization of a User Model (UM) is im portant for every system that provides persona... more Accurate initialization of a User Model (UM) is im portant for every system that provides personalized services. Howeve r, th re are systems where this initialization is critical yet no user data is available from prior interactions. For example, one in which both the repeat usage by a user is rare, and the total interaction on a single use is relatively limited. Evaluating the effectiveness of a User Model and particular instance of such a model is never an easy task, particularly through ser studies. Moreover, evaluation typically focuses on the usability of an entire system rather than the performance of a specific UM instantiation. In thi s paper we propose evaluating the quality of UMs via simulation and co mparison to a "gold standard". This standard is an approximation of th e user's ideal model. We will demonstrate this through a case study of a museum's visitor guide system implemented in the Hecht Museum at the University o f Haifa, Israel
We show how one-class compression Neural Networks and one-class SVM can be applied to fMRI data t... more We show how one-class compression Neural Networks and one-class SVM can be applied to fMRI data to learn the classification of brain activity associated with a specific motor activity. For comparison purposes, we use two labeled data and see what degree of classification ability is lost compared with the usual two-class SVM.
2020 International Joint Conference on Neural Networks (IJCNN), 2020
We develop a method that is based on processing gathered Event Related Potentials (ERP) signals a... more We develop a method that is based on processing gathered Event Related Potentials (ERP) signals and the use of machine learning technique for multivariate analysis (i.e. classification) that we apply in order to analyze the differences between Dyslexic and Skilled readers.No human intervention is needed in the analysis process. This is the state of the art results for automatic identification of Dyslexic readers using a Lexical Decision Task. We use mathematical and machine learning based techniques to automatically discover novel complex features that (i) allow for reliable distinction between Dyslexic and Normal Control Skilled readers and (ii) to validate the assumption that most of the differences between Dyslexic and Skilled readers are located in the left hemisphere.Interestingly, these tools also pointed to the fact that High Pass signals (typically considered as "noise" during ERP/EEG analyses) in fact contain significant relevant information.Finally, the proposed ...
We implemented versions of the SVM appropriate for one-class classification in the context of inf... more We implemented versions of the SVM appropriate for one-class classification in the context of information retrieval. The experiments were conducted on the standard Reuters data set. For the SVM implementation we used both a version of Schoelkopf et al. and a somewhat different version of one-class SVM based on identifying "outlier" data as representative of the second-class. We report on experiments with different kernels for both of these implementations and with different representations of the data, including binary vectors, tf-idf representation and a modification called "Hadamard" representation. Then we compared it with one-class versions of the algorithms prototype (Rocchio), nearest neighbor, naive Bayes, and finally a natural one-class neural network classification method based on "bottleneck" compression generated filters.The SVM approach as represented by Schoelkopf was superior to all the methods except the neural network one, where it was, ...
We develop a method that is based on processing gathered Event Related Potentials (ERP) signals a... more We develop a method that is based on processing gathered Event Related Potentials (ERP) signals and the use of machine learning technique for multivariate analysis (i.e. classification) that we apply in order to analyze the differences between Dyslexic and Skilled readers. No human intervention is needed in the analysis process. This is the state of the art results for automatic identification of Dyslexic readers using a Lexical Decision Task. We use mathematical and machine learning based techniques to automatically discover novel complex features that (i) allow for reliable distinction between Dyslexic and Normal Control Skilled readers and (ii) to validate the assumption that the most of the differences between Dyslexic and Skilled readers located in the left hemisphere. Interestingly, these tools also pointed to the fact that High Pass signals (typically considered as "noise" during ERP/EEG analyses) in fact contains significant relevant information. Finally, the propo...
Recent advances in the DNA microarray technology have provided the ability to examine and measure... more Recent advances in the DNA microarray technology have provided the ability to examine and measure the expression levels of thousands of genes simultaneously in an organism. In this technology each gene is recorded under different conditions or each gene is evaluated under a single environment but in different types of tissues. In the first case it is used in identification of functionally related genes where asthe second type of technology is helpful in classification of different types of tissues and identification of those genes whose expression levels are good diagnostic indicators. Different approaches have been applied to classify different datasets. However, the main challenges in this task is the availability of a smaller number of samples compared to huge number of genes and the noisy nature of biological data. This paper review on different techniques used to classify the genes and improved efficiency of biomarker identification due to these classifications.
Using Machine Learning Tools (Neural Networks and Support Vector Machines) we show how raw fMRI b... more Using Machine Learning Tools (Neural Networks and Support Vector Machines) we show how raw fMRI brain scan data can be correctly assigned to cognitive tasks. We describe experiments classifying visual and motor tasks using one-class and two-class labeling for training. No a priori knowledge (e.g. of anatomy or physiology) is needed for the system to work. These results further suggest that feature reduction techniques may allow for the automatic location of brain areas correlated with specific cognitive tasks possibly evn when the needed features are not localized.
Uploads
Papers by Larry Manevitz