Pub Date : 2015-07-12DOI: 10.1109/IJCNN.2015.7280646
Shibani Santurkar, B. Rajendran
We demonstrate a spiking neural network for navigation motivated by the chemotaxis circuit of Caenorhabditis elegans. Our network uses information regarding temporal gradients in intensity of local variables such as chemical concentration, temperature, radiation, etc., to make navigational decisions for contour tracking and obstacle avoidance. The gradient information is determined by mimicking the underlying mechanisms of the ASE neurons of C. elegans. Simulations show that our software-worm is able to identify the set-point with 92% efficiency, 68.5% higher than an optimal memoryless Lévy foraging strategy and 33% higher than an equivalent non-spiking neural network configuration. The software-worm is able to track the set-point with an average deviation of 1% from the set-point, and this performance degrades merely by 1.8% in the presence of intense salt and pepper noise in the local tracking variable. We also develop a VLSI implementation for the main gradient detector neurons, which could be integrated with standard comparator circuitry to develop robust circuits for navigation and contour tracking. We demonstrate noise-resilience of our network to environmental, architectural and circuit noise.
{"title":"C. elegans chemotaxis inspired neuromorphic circuit for contour tracking and obstacle avoidance","authors":"Shibani Santurkar, B. Rajendran","doi":"10.1109/IJCNN.2015.7280646","DOIUrl":"https://doi.org/10.1109/IJCNN.2015.7280646","url":null,"abstract":"We demonstrate a spiking neural network for navigation motivated by the chemotaxis circuit of Caenorhabditis elegans. Our network uses information regarding temporal gradients in intensity of local variables such as chemical concentration, temperature, radiation, etc., to make navigational decisions for contour tracking and obstacle avoidance. The gradient information is determined by mimicking the underlying mechanisms of the ASE neurons of C. elegans. Simulations show that our software-worm is able to identify the set-point with 92% efficiency, 68.5% higher than an optimal memoryless Lévy foraging strategy and 33% higher than an equivalent non-spiking neural network configuration. The software-worm is able to track the set-point with an average deviation of 1% from the set-point, and this performance degrades merely by 1.8% in the presence of intense salt and pepper noise in the local tracking variable. We also develop a VLSI implementation for the main gradient detector neurons, which could be integrated with standard comparator circuitry to develop robust circuits for navigation and contour tracking. We demonstrate noise-resilience of our network to environmental, architectural and circuit noise.","PeriodicalId":6539,"journal":{"name":"2015 International Joint Conference on Neural Networks (IJCNN)","volume":"8 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2015-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89477700","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 : 2015-07-12DOI: 10.1109/IJCNN.2015.7280597
Ken-ichiro Nishioka, Yoshitatsu Matsuda, K. Yamaguchi
In this paper, a new approach is proposed for extracting localized spatio-temporal patterns from Foursquare, which is a location-based social networking system (SNS). Previously, we have proposed a method estimating the probabilistic distribution of users in Foursquare by a diffusion-type formula and have extracted various spatio-temporal patterns from the distribution by principal component analysis. However, as the distribution was the average over all the users, only the “global” patterns were extracted. So, we can not extract localized patterns showing the detailed behaviors of limited users in local areas. In this paper, a new method is proposed in order to extract the localized patterns by clustering users. First, the distance among users is measured by the Hellinger distance among the distributions of each user. Next, Ward's method (which is a widely used method in hierarchical cluster analysis) is applied to the users with their distance. Finally, the spatio-temporal patterns are extracted from the distributions for each cluster of users. The results on the real Foursquare dataset show that the proposed method can extract various and interesting localized patterns from each cluster of users.
{"title":"Discovery of localized spatio-temporal patterns from location-based SNS by clustering users","authors":"Ken-ichiro Nishioka, Yoshitatsu Matsuda, K. Yamaguchi","doi":"10.1109/IJCNN.2015.7280597","DOIUrl":"https://doi.org/10.1109/IJCNN.2015.7280597","url":null,"abstract":"In this paper, a new approach is proposed for extracting localized spatio-temporal patterns from Foursquare, which is a location-based social networking system (SNS). Previously, we have proposed a method estimating the probabilistic distribution of users in Foursquare by a diffusion-type formula and have extracted various spatio-temporal patterns from the distribution by principal component analysis. However, as the distribution was the average over all the users, only the “global” patterns were extracted. So, we can not extract localized patterns showing the detailed behaviors of limited users in local areas. In this paper, a new method is proposed in order to extract the localized patterns by clustering users. First, the distance among users is measured by the Hellinger distance among the distributions of each user. Next, Ward's method (which is a widely used method in hierarchical cluster analysis) is applied to the users with their distance. Finally, the spatio-temporal patterns are extracted from the distributions for each cluster of users. The results on the real Foursquare dataset show that the proposed method can extract various and interesting localized patterns from each cluster of users.","PeriodicalId":6539,"journal":{"name":"2015 International Joint Conference on Neural Networks (IJCNN)","volume":"14 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2015-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89923859","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 : 2015-07-12DOI: 10.1109/IJCNN.2015.7280558
Luiz G. Hafemann, Luiz Oliveira, P. Cavalin, R. Sabourin
Convolutional Neural Networks (CNNs) have set the state-of-the-art in many computer vision tasks in recent years. For this type of model, it is common to have millions of parameters to train, commonly requiring large datasets. We investigate a method to transfer learning across different texture classification problems, using CNNs, in order to take advantage of this type of architecture to problems with smaller datasets. We use a Convolutional Neural Network trained on a source dataset (with lots of data) to project the data of a target dataset (with limited data) onto another feature space, and then train a classifier on top of this new representation. Our experiments show that this technique can achieve good results in tasks with small datasets, by leveraging knowledge learned from tasks with larger datasets. Testing the method on the the Brodatz-32 dataset, we achieved an accuracy of 97.04% - superior to models trained with popular texture descriptors, such as Local Binary Patterns and Gabor Filters, and increasing the accuracy by 6 percentage points compared to a CNN trained directly on the Brodatz-32 dataset. We also present a visual analysis of the projected dataset, showing that the data is projected to a space where samples from the same class are clustered together - suggesting that the features learned by the CNN in the source task are relevant for the target task.
{"title":"Transfer learning between texture classification tasks using Convolutional Neural Networks","authors":"Luiz G. Hafemann, Luiz Oliveira, P. Cavalin, R. Sabourin","doi":"10.1109/IJCNN.2015.7280558","DOIUrl":"https://doi.org/10.1109/IJCNN.2015.7280558","url":null,"abstract":"Convolutional Neural Networks (CNNs) have set the state-of-the-art in many computer vision tasks in recent years. For this type of model, it is common to have millions of parameters to train, commonly requiring large datasets. We investigate a method to transfer learning across different texture classification problems, using CNNs, in order to take advantage of this type of architecture to problems with smaller datasets. We use a Convolutional Neural Network trained on a source dataset (with lots of data) to project the data of a target dataset (with limited data) onto another feature space, and then train a classifier on top of this new representation. Our experiments show that this technique can achieve good results in tasks with small datasets, by leveraging knowledge learned from tasks with larger datasets. Testing the method on the the Brodatz-32 dataset, we achieved an accuracy of 97.04% - superior to models trained with popular texture descriptors, such as Local Binary Patterns and Gabor Filters, and increasing the accuracy by 6 percentage points compared to a CNN trained directly on the Brodatz-32 dataset. We also present a visual analysis of the projected dataset, showing that the data is projected to a space where samples from the same class are clustered together - suggesting that the features learned by the CNN in the source task are relevant for the target task.","PeriodicalId":6539,"journal":{"name":"2015 International Joint Conference on Neural Networks (IJCNN)","volume":"80 1","pages":"1-7"},"PeriodicalIF":0.0,"publicationDate":"2015-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90924436","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 : 2015-07-12DOI: 10.1109/IJCNN.2015.7280686
E. Daucé, T. Proix, L. Ralaivola
We adapt a policy gradient approach to the problem of reward-based online learning of a non-invasive EEG-based “P300”-speller. We first clarify the nature of the P300-speller classification problem and present a general regularized gradient ascent formula. We then show that when the reward is immediate and binary (namely “bad response” or “good response”), each update is expected to improve the classifier accuracy, whether the actual response is correct or not. We also estimate the robustness of the method to occasional mistaken rewards, i.e. show that the learning efficacy may only linearly decrease with the rate of invalid rewards. The effectiveness of our approach is tested in a series of simulations reproducing the conditions of real experiments. We show in a first experiment that a systematic improvement of the spelling rate is obtained for all subjects in the absence of initial calibration. In a second experiment, we consider the case of the online recovery that is expected to follow failed electrodes. Combined with a specific failure detection algorithm, the spelling error information (typically contained in a “backspace” hit) is shown useful for the policy gradient to adapt the P300 classifier to the new situation, provided the feedback is reliable enough (namely having a reliability greater than 70%).
{"title":"Reward-based online learning in non-stationary environments: Adapting a P300-speller with a “backspace” key","authors":"E. Daucé, T. Proix, L. Ralaivola","doi":"10.1109/IJCNN.2015.7280686","DOIUrl":"https://doi.org/10.1109/IJCNN.2015.7280686","url":null,"abstract":"We adapt a policy gradient approach to the problem of reward-based online learning of a non-invasive EEG-based “P300”-speller. We first clarify the nature of the P300-speller classification problem and present a general regularized gradient ascent formula. We then show that when the reward is immediate and binary (namely “bad response” or “good response”), each update is expected to improve the classifier accuracy, whether the actual response is correct or not. We also estimate the robustness of the method to occasional mistaken rewards, i.e. show that the learning efficacy may only linearly decrease with the rate of invalid rewards. The effectiveness of our approach is tested in a series of simulations reproducing the conditions of real experiments. We show in a first experiment that a systematic improvement of the spelling rate is obtained for all subjects in the absence of initial calibration. In a second experiment, we consider the case of the online recovery that is expected to follow failed electrodes. Combined with a specific failure detection algorithm, the spelling error information (typically contained in a “backspace” hit) is shown useful for the policy gradient to adapt the P300 classifier to the new situation, provided the feedback is reliable enough (namely having a reliability greater than 70%).","PeriodicalId":6539,"journal":{"name":"2015 International Joint Conference on Neural Networks (IJCNN)","volume":"54 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2015-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90935640","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 : 2015-07-12DOI: 10.1109/IJCNN.2015.7280715
Amirhosein Shantia, Rik Timmers, Lambert Schomaker, M. Wiering
Robotic mapping and localization methods are mostly dominated by using a combination of spatial alignment of sensory inputs, loop closure detection, and a global fine-tuning step. This requires either expensive depth sensing systems, or fast computational hardware at run-time to produce a 2D or 3D map of the environment. In a similar context, deep neural networks are used extensively in scene recognition applications, but are not yet applied to localization and mapping problems. In this paper, we adopt a novel approach by using denoising autoencoders and image information for tackling robot localization problems. We use semi-supervised learning with location values that are provided by traditional mapping methods. After training, our method requires much less run-time computations, and therefore can perform real-time localization on normal processing units. We compare the effects of different feature vectors such as plain images, the scale invariant feature transform and histograms of oriented gradients on the localization precision. The best system can localize with an average positional error of ten centimeters and an angular error of four degrees in 3D simulation.
{"title":"Indoor localization by denoising autoencoders and semi-supervised learning in 3D simulated environment","authors":"Amirhosein Shantia, Rik Timmers, Lambert Schomaker, M. Wiering","doi":"10.1109/IJCNN.2015.7280715","DOIUrl":"https://doi.org/10.1109/IJCNN.2015.7280715","url":null,"abstract":"Robotic mapping and localization methods are mostly dominated by using a combination of spatial alignment of sensory inputs, loop closure detection, and a global fine-tuning step. This requires either expensive depth sensing systems, or fast computational hardware at run-time to produce a 2D or 3D map of the environment. In a similar context, deep neural networks are used extensively in scene recognition applications, but are not yet applied to localization and mapping problems. In this paper, we adopt a novel approach by using denoising autoencoders and image information for tackling robot localization problems. We use semi-supervised learning with location values that are provided by traditional mapping methods. After training, our method requires much less run-time computations, and therefore can perform real-time localization on normal processing units. We compare the effects of different feature vectors such as plain images, the scale invariant feature transform and histograms of oriented gradients on the localization precision. The best system can localize with an average positional error of ten centimeters and an angular error of four degrees in 3D simulation.","PeriodicalId":6539,"journal":{"name":"2015 International Joint Conference on Neural Networks (IJCNN)","volume":"76 1","pages":"1-7"},"PeriodicalIF":0.0,"publicationDate":"2015-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"91186731","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 : 2015-07-12DOI: 10.1109/IJCNN.2015.7280834
Xiaofeng Shi, Guoqiang Xu, S. Furao, Jinxi Zhao
The imbalance problem exists in P300 EEG data sets because P300 potential are collected under the condition of Oddball experimental paradigm. Hence, a P300 detection method, namely RUSBagging SVMs, is proposed in this paper to solve the imbalance problem and make an improvement. This algorithm re-samples the data sets at first to generate a rebalanced training set in one round of iteration and trains an SVM classifier based on the training set. Next, the SVM classifiers are integrated to make a final decision. In the integration of several classifiers, the information that is lost in the under-sampling process is generally considered. Therefore, the method is relatively robust. The experiments of character recognition based on P300 EEG data signals are conducted to examine the method. It is concluded from the experiments that RUSBagging method can indeed improve the performance of P300 detection by solving the imbalance problem in EEG data sets.
{"title":"Solving the data imbalance problem of P300 detection via Random Under-Sampling Bagging SVMs","authors":"Xiaofeng Shi, Guoqiang Xu, S. Furao, Jinxi Zhao","doi":"10.1109/IJCNN.2015.7280834","DOIUrl":"https://doi.org/10.1109/IJCNN.2015.7280834","url":null,"abstract":"The imbalance problem exists in P300 EEG data sets because P300 potential are collected under the condition of Oddball experimental paradigm. Hence, a P300 detection method, namely RUSBagging SVMs, is proposed in this paper to solve the imbalance problem and make an improvement. This algorithm re-samples the data sets at first to generate a rebalanced training set in one round of iteration and trains an SVM classifier based on the training set. Next, the SVM classifiers are integrated to make a final decision. In the integration of several classifiers, the information that is lost in the under-sampling process is generally considered. Therefore, the method is relatively robust. The experiments of character recognition based on P300 EEG data signals are conducted to examine the method. It is concluded from the experiments that RUSBagging method can indeed improve the performance of P300 detection by solving the imbalance problem in EEG data sets.","PeriodicalId":6539,"journal":{"name":"2015 International Joint Conference on Neural Networks (IJCNN)","volume":"12 1","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2015-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"91364347","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 : 2015-07-12DOI: 10.1109/IJCNN.2015.7280372
Xiang Xu, F. Lin, Carol Ng, K. Leong
Indirect Immunofluorescence (IIF) on Human Epithelial-2 (HEp-2) cells is the hallmark method for detecting some specific autoimmune diseases by identifying the presence of antinuclear antibodies (ANAs) within a patient's serum. Due to the limitations of IIF, such as being subjective and time consuming, automated Computer-aided diagnosis (CAD) system is required for diagnostic purposes. In this paper, we propose a novel feature extraction scheme for automatic staining pattern classification of HEp-2 cells. Our method constructs a dual spatial pyramid structure on a powerful rotation invariant texture feature, which has the following advantages: (1) invariance under local rotation of the image, (2) robustness against resolution changes, and (3) strong descriptive ability. Incorporated with a linear SVM classifier, our approach demonstrates its effectiveness by testing on two HEp-2 cells datasets: the ICPR2012 dataset and the ICIP2013 training dataset. Particularly, it shows superior classification performance than the best performer at the first edition of the HEp-2 cell classification contest.
{"title":"Dual Spatial Pyramid On Rotation Invariant Texture Feature For HEp-2 Cell Classification","authors":"Xiang Xu, F. Lin, Carol Ng, K. Leong","doi":"10.1109/IJCNN.2015.7280372","DOIUrl":"https://doi.org/10.1109/IJCNN.2015.7280372","url":null,"abstract":"Indirect Immunofluorescence (IIF) on Human Epithelial-2 (HEp-2) cells is the hallmark method for detecting some specific autoimmune diseases by identifying the presence of antinuclear antibodies (ANAs) within a patient's serum. Due to the limitations of IIF, such as being subjective and time consuming, automated Computer-aided diagnosis (CAD) system is required for diagnostic purposes. In this paper, we propose a novel feature extraction scheme for automatic staining pattern classification of HEp-2 cells. Our method constructs a dual spatial pyramid structure on a powerful rotation invariant texture feature, which has the following advantages: (1) invariance under local rotation of the image, (2) robustness against resolution changes, and (3) strong descriptive ability. Incorporated with a linear SVM classifier, our approach demonstrates its effectiveness by testing on two HEp-2 cells datasets: the ICPR2012 dataset and the ICIP2013 training dataset. Particularly, it shows superior classification performance than the best performer at the first edition of the HEp-2 cell classification contest.","PeriodicalId":6539,"journal":{"name":"2015 International Joint Conference on Neural Networks (IJCNN)","volume":"24 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2015-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89702884","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 : 2015-07-12DOI: 10.1109/IJCNN.2015.7280737
Haider Raza, H. Cecotti, G. Prasad
A major problem in a brain-computer interface (BCI) based on electroencephalogram (EEG) recordings is the varying statistical properties of the signals during inter- or intra-session transfers that often lead to deteriorated BCI performances. A filter bank CSP (FBCSP) algorithm typically uses all the features from all the bands to extract and select robust features. In this paper, we evaluate the performance of four methods for frequency band selection applied to binary motor imagery classification: forward-addition (FA), backward-elimination (BE), the intersection and the union of the FA and BE. These methods automatically select and learn the best discriminative sets of frequency bands, and their corresponding CSP features. The performances of the proposed methods are evaluated on binary motor imagery classification using a publicly available real-world dataset (BCI competition 2008 dataset 2A). It is found that the BE method provides the best improvement resulting in an average classification accuracy increase of the BCI system over the FBCSP algorithm, from 77.06% to 79.09%.
{"title":"Optimising frequency band selection with forward-addition and backward-elimination algorithms in EEG-based brain-computer interfaces","authors":"Haider Raza, H. Cecotti, G. Prasad","doi":"10.1109/IJCNN.2015.7280737","DOIUrl":"https://doi.org/10.1109/IJCNN.2015.7280737","url":null,"abstract":"A major problem in a brain-computer interface (BCI) based on electroencephalogram (EEG) recordings is the varying statistical properties of the signals during inter- or intra-session transfers that often lead to deteriorated BCI performances. A filter bank CSP (FBCSP) algorithm typically uses all the features from all the bands to extract and select robust features. In this paper, we evaluate the performance of four methods for frequency band selection applied to binary motor imagery classification: forward-addition (FA), backward-elimination (BE), the intersection and the union of the FA and BE. These methods automatically select and learn the best discriminative sets of frequency bands, and their corresponding CSP features. The performances of the proposed methods are evaluated on binary motor imagery classification using a publicly available real-world dataset (BCI competition 2008 dataset 2A). It is found that the BE method provides the best improvement resulting in an average classification accuracy increase of the BCI system over the FBCSP algorithm, from 77.06% to 79.09%.","PeriodicalId":6539,"journal":{"name":"2015 International Joint Conference on Neural Networks (IJCNN)","volume":"38 1","pages":"1-7"},"PeriodicalIF":0.0,"publicationDate":"2015-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90398022","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 : 2015-07-12DOI: 10.1109/IJCNN.2015.7280773
Madhavun Candadai, A. Vanarase, M. Mei, A. Minai
The availability of unstructured text as a source of data has increased by orders of magnitude in the last few years, triggering extensive research in the automated processing and analysis of electronic texts. An especially important and difficult problem is the identification of salient words in a corpus, so that further processing can focus on these words without distraction by uninformative words. Standard lists of stop words are used to remove common words such as articles, pronouns and prepositions, but many other words that should be removed are much harder to identify because word salience is highly context-dependent. In this paper, we describe a neurodynamical approach for the context-dependent identification of salient words in large text corpora. The method, termed the Attractor Network-based Salient Word Extraction Rule (ANSWER) is modeled as a cognitive mechanism that identifies salient words based on their participation in coherent multi-word ideas. These ideas are, in turn, extracted via attractor dynamics in a recurrent neural network modeling the associative semantic graph of the corpus. The corpus used in this paper comprises the abstracts of all papers published in the proceedings of IJCNN 2009, 2011 and 2013. The list of salient words that the system generates is compared with those generated by other standard metrics, and is found to outperform all of them in almost all cases.
{"title":"ANSWER: An unsupervised attractor network method for detecting salient words in text corpora","authors":"Madhavun Candadai, A. Vanarase, M. Mei, A. Minai","doi":"10.1109/IJCNN.2015.7280773","DOIUrl":"https://doi.org/10.1109/IJCNN.2015.7280773","url":null,"abstract":"The availability of unstructured text as a source of data has increased by orders of magnitude in the last few years, triggering extensive research in the automated processing and analysis of electronic texts. An especially important and difficult problem is the identification of salient words in a corpus, so that further processing can focus on these words without distraction by uninformative words. Standard lists of stop words are used to remove common words such as articles, pronouns and prepositions, but many other words that should be removed are much harder to identify because word salience is highly context-dependent. In this paper, we describe a neurodynamical approach for the context-dependent identification of salient words in large text corpora. The method, termed the Attractor Network-based Salient Word Extraction Rule (ANSWER) is modeled as a cognitive mechanism that identifies salient words based on their participation in coherent multi-word ideas. These ideas are, in turn, extracted via attractor dynamics in a recurrent neural network modeling the associative semantic graph of the corpus. The corpus used in this paper comprises the abstracts of all papers published in the proceedings of IJCNN 2009, 2011 and 2013. The list of salient words that the system generates is compared with those generated by other standard metrics, and is found to outperform all of them in almost all cases.","PeriodicalId":6539,"journal":{"name":"2015 International Joint Conference on Neural Networks (IJCNN)","volume":"62 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2015-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78551804","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 : 2015-07-12DOI: 10.1109/IJCNN.2015.7280728
Eugene Koskin, D. Galayko, O. Feely, E. Blokhina
In this paper we consider a network of phase oscillators. We develop the equations that model the time evolution of the phase of each oscillator in the network. The oscillator represents a modified Kuramoto oscillator and in this study we discuss how these modifications are obtained. In the context of this study, we use this network to model a network of PLLs for distributed clock applications. We analyse analytically and numerically the synchronisation modes of this system for different types of the coupling function. We show that depending on the properties of the coupling function, the network displays either multiple coexisting synchronisation modes or only a single synchronisation mode. While in the context of clock generation, multiple synchronisation modes coexisting in the system at the same parameters are a parasitic phenomenon. However in the context of other application such as associative memory models, mode-locking can be seen a useful phenomenon. The results provide a deeper understanding of globally synchronised clock networks with applications in microprocessor design.
{"title":"Mode-locking in a network of kuramoto-like oscillators","authors":"Eugene Koskin, D. Galayko, O. Feely, E. Blokhina","doi":"10.1109/IJCNN.2015.7280728","DOIUrl":"https://doi.org/10.1109/IJCNN.2015.7280728","url":null,"abstract":"In this paper we consider a network of phase oscillators. We develop the equations that model the time evolution of the phase of each oscillator in the network. The oscillator represents a modified Kuramoto oscillator and in this study we discuss how these modifications are obtained. In the context of this study, we use this network to model a network of PLLs for distributed clock applications. We analyse analytically and numerically the synchronisation modes of this system for different types of the coupling function. We show that depending on the properties of the coupling function, the network displays either multiple coexisting synchronisation modes or only a single synchronisation mode. While in the context of clock generation, multiple synchronisation modes coexisting in the system at the same parameters are a parasitic phenomenon. However in the context of other application such as associative memory models, mode-locking can be seen a useful phenomenon. The results provide a deeper understanding of globally synchronised clock networks with applications in microprocessor design.","PeriodicalId":6539,"journal":{"name":"2015 International Joint Conference on Neural Networks (IJCNN)","volume":"20 1","pages":"1-6"},"PeriodicalIF":0.0,"publicationDate":"2015-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78577164","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：