Pub Date : 2013-12-01DOI: 10.1109/IJCNN.2013.6706861
Benhui Chen, Xuefen Hong, Lihua Duan, Jinglu Hu
Multi-label classification is an extension of traditional classification problem in which each instance is associated with a set of labels. For some multi-label classification tasks, labels are usually overlapped and correlated, and some implicit constraint rules are existed among the labels. This paper presents an improved multi-label classification method based on label ranking strategy and label constraints. Firstly, one-against-all decomposition technique is used to divide a multilabel classification task into multiple independent binary classification sub-problems. One binary SVM classifier is trained for each label. Secondly, based on training data, label constraint rules are mined by association rule learning method. Thirdly, a correction model based on label constraints is used to correct the probabilistic outputs of SVM classifiers for label ranking. Experiment results on three well-known multi-label benchmark datasets show that the proposed method outperforms some conventional multi-label classification methods.
{"title":"Improving multi-label classification performance by label constraints","authors":"Benhui Chen, Xuefen Hong, Lihua Duan, Jinglu Hu","doi":"10.1109/IJCNN.2013.6706861","DOIUrl":"https://doi.org/10.1109/IJCNN.2013.6706861","url":null,"abstract":"Multi-label classification is an extension of traditional classification problem in which each instance is associated with a set of labels. For some multi-label classification tasks, labels are usually overlapped and correlated, and some implicit constraint rules are existed among the labels. This paper presents an improved multi-label classification method based on label ranking strategy and label constraints. Firstly, one-against-all decomposition technique is used to divide a multilabel classification task into multiple independent binary classification sub-problems. One binary SVM classifier is trained for each label. Secondly, based on training data, label constraint rules are mined by association rule learning method. Thirdly, a correction model based on label constraints is used to correct the probabilistic outputs of SVM classifiers for label ranking. Experiment results on three well-known multi-label benchmark datasets show that the proposed method outperforms some conventional multi-label classification methods.","PeriodicalId":376975,"journal":{"name":"The 2013 International Joint Conference on Neural Networks (IJCNN)","volume":"543 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123266704","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 : 2013-12-01DOI: 10.1109/IJCNN.2013.6707140
D. Reid, A. Hussain, H. Tawfik
In this paper a novel application of a particular type of spiking neural network, a Polychronous Spiking Network, for financial time series prediction is introduced with the aim of exploiting the inherent temporal capabilities of the spiking neural model. The performance of the spiking neural network was benchmarked against two “traditional”, rate-encoded, neural networks; a Multi-Layer Perceptron network and a Functional Link Neural Network. Three nonstationary and noisy time series are used to test these simulations: IBM stock data; US/Euro exchange rate data, and the price of Brent crude oil. The experiments demonstrated favourable prediction results for the Spiking Neural Network in terms of Annualised Return, for both 1-Step and 5-Step ahead predictions. These results were also supported by other relevant metrics such as Maximum Drawdown, Signal-To-Noise ratio, and Normalised Mean Square Error. This work demonstrated the applicability of polychronous spiking network to financial data forecasting and that it has the potential to function more effectively than traditional neural networks, in nonstationary environments.
{"title":"Spiking neural networks for financial data prediction","authors":"D. Reid, A. Hussain, H. Tawfik","doi":"10.1109/IJCNN.2013.6707140","DOIUrl":"https://doi.org/10.1109/IJCNN.2013.6707140","url":null,"abstract":"In this paper a novel application of a particular type of spiking neural network, a Polychronous Spiking Network, for financial time series prediction is introduced with the aim of exploiting the inherent temporal capabilities of the spiking neural model. The performance of the spiking neural network was benchmarked against two “traditional”, rate-encoded, neural networks; a Multi-Layer Perceptron network and a Functional Link Neural Network. Three nonstationary and noisy time series are used to test these simulations: IBM stock data; US/Euro exchange rate data, and the price of Brent crude oil. The experiments demonstrated favourable prediction results for the Spiking Neural Network in terms of Annualised Return, for both 1-Step and 5-Step ahead predictions. These results were also supported by other relevant metrics such as Maximum Drawdown, Signal-To-Noise ratio, and Normalised Mean Square Error. This work demonstrated the applicability of polychronous spiking network to financial data forecasting and that it has the potential to function more effectively than traditional neural networks, in nonstationary environments.","PeriodicalId":376975,"journal":{"name":"The 2013 International Joint Conference on Neural Networks (IJCNN)","volume":"95 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123227616","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 : 2013-12-01DOI: 10.1109/IJCNN.2013.6706743
Yuling Lin, Haixiang Guo, Jinglu Hu
In this paper, an SVM-based approach is proposed for stock market trend prediction. The proposed approach consists of two parts: feature selection and prediction model. In the feature selection part, a correlation-based SVM filter is applied to rank and select a good subset of financial indexes. And the stock indicators are evaluated based on the ranking. In the prediction model part, a so called quasi-linear SVM is applied to predict stock market movement direction in term of historical data series by using the selected subset of financial indexes as the weighted inputs. The quasi-linear SVM is an SVM with a composite quasi-linear kernel function, which approximates a nonlinear separating boundary by multi-local linear classifiers with interpolation. Experimental results on Taiwan stock market datasets demonstrate that the proposed SVM-based stock market trend prediction method produces better generalization performance over the conventional methods in terms of the hit ratio. Moreover, the experimental results also show that the proposed SVM-based stock market trend prediction system can find out a good subset and evaluate stock indicators which provide useful information for investors.
{"title":"An SVM-based approach for stock market trend prediction","authors":"Yuling Lin, Haixiang Guo, Jinglu Hu","doi":"10.1109/IJCNN.2013.6706743","DOIUrl":"https://doi.org/10.1109/IJCNN.2013.6706743","url":null,"abstract":"In this paper, an SVM-based approach is proposed for stock market trend prediction. The proposed approach consists of two parts: feature selection and prediction model. In the feature selection part, a correlation-based SVM filter is applied to rank and select a good subset of financial indexes. And the stock indicators are evaluated based on the ranking. In the prediction model part, a so called quasi-linear SVM is applied to predict stock market movement direction in term of historical data series by using the selected subset of financial indexes as the weighted inputs. The quasi-linear SVM is an SVM with a composite quasi-linear kernel function, which approximates a nonlinear separating boundary by multi-local linear classifiers with interpolation. Experimental results on Taiwan stock market datasets demonstrate that the proposed SVM-based stock market trend prediction method produces better generalization performance over the conventional methods in terms of the hit ratio. Moreover, the experimental results also show that the proposed SVM-based stock market trend prediction system can find out a good subset and evaluate stock indicators which provide useful information for investors.","PeriodicalId":376975,"journal":{"name":"The 2013 International Joint Conference on Neural Networks (IJCNN)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121475530","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 : 2013-08-08DOI: 10.1109/IJCNN.2013.6706968
D. Kerr, S. Coleman, T. McGinnity, Marine Clogenson
The recent development of low cost cameras that capture 3-dimensional images has changed the focus of computer vision research from using solely intensity images to the use of range images, or combinations of RGB, intensity and range images. The low cost and widespread availability of the hardware to capture these images has realised many possible applications in areas such as robotics, object recognition, surveillance, manipulation, navigation and interaction. Given the large volumes of data in range images, processing and extracting the relevant information from the images in real time becomes challenging. To achieve this, much research has been conducted in the area of bio-inspired feature extraction which aims to emulate the biological processes used to extract relevant features, reduce redundancy, and process images efficiently. Inspired by the behaviour of biological vision systems, an approach is presented for extracting important features from intensity and range images, using biologically inspired spiking neural networks in order to model aspects of the functional computational capabilities of the visual system.
{"title":"Biologically inspired intensity and range image feature extraction","authors":"D. Kerr, S. Coleman, T. McGinnity, Marine Clogenson","doi":"10.1109/IJCNN.2013.6706968","DOIUrl":"https://doi.org/10.1109/IJCNN.2013.6706968","url":null,"abstract":"The recent development of low cost cameras that capture 3-dimensional images has changed the focus of computer vision research from using solely intensity images to the use of range images, or combinations of RGB, intensity and range images. The low cost and widespread availability of the hardware to capture these images has realised many possible applications in areas such as robotics, object recognition, surveillance, manipulation, navigation and interaction. Given the large volumes of data in range images, processing and extracting the relevant information from the images in real time becomes challenging. To achieve this, much research has been conducted in the area of bio-inspired feature extraction which aims to emulate the biological processes used to extract relevant features, reduce redundancy, and process images efficiently. Inspired by the behaviour of biological vision systems, an approach is presented for extracting important features from intensity and range images, using biologically inspired spiking neural networks in order to model aspects of the functional computational capabilities of the visual system.","PeriodicalId":376975,"journal":{"name":"The 2013 International Joint Conference on Neural Networks (IJCNN)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-08-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122679400","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 : 2013-08-04DOI: 10.1109/IJCNN.2013.6706927
Evangelos Stromatias, F. Galluppi, Cameron Patterson, S. Furber
Simulating large spiking neural networks is non trivial: supercomputers offer great flexibility at the price of power and communication overheads; custom neuromorphic circuits are more power efficient but less flexible; while alternative approaches based on GPGPUs and FPGAs, whilst being more readily available, show similar model specialization. As well as efficiency and flexibility, real time simulation is a desirable neural network characteristic, for example in cognitive robotics where embodied agents interact with the environment using low-power, event-based neuromorphic sensors. The SpiNNaker neuromimetic architecture has been designed to address these requirements, simulating large-scale heterogeneous models of spiking neurons in real-time, offering a unique combination of flexibility, scalability and power efficiency. In this work a 48-chip board is utilised to generate a SpiNNaker power estimation model, based on numbers of neurons, synapses and their firing rates. In addition, we demonstrate simulations capable of handling up to a quarter of a million neurons, 81 million synapses and 1.8 billion synaptic events per second, with the most complex simulations consuming less than 1 Watt per SpiNNaker chip.
{"title":"Power analysis of large-scale, real-time neural networks on SpiNNaker","authors":"Evangelos Stromatias, F. Galluppi, Cameron Patterson, S. Furber","doi":"10.1109/IJCNN.2013.6706927","DOIUrl":"https://doi.org/10.1109/IJCNN.2013.6706927","url":null,"abstract":"Simulating large spiking neural networks is non trivial: supercomputers offer great flexibility at the price of power and communication overheads; custom neuromorphic circuits are more power efficient but less flexible; while alternative approaches based on GPGPUs and FPGAs, whilst being more readily available, show similar model specialization. As well as efficiency and flexibility, real time simulation is a desirable neural network characteristic, for example in cognitive robotics where embodied agents interact with the environment using low-power, event-based neuromorphic sensors. The SpiNNaker neuromimetic architecture has been designed to address these requirements, simulating large-scale heterogeneous models of spiking neurons in real-time, offering a unique combination of flexibility, scalability and power efficiency. In this work a 48-chip board is utilised to generate a SpiNNaker power estimation model, based on numbers of neurons, synapses and their firing rates. In addition, we demonstrate simulations capable of handling up to a quarter of a million neurons, 81 million synapses and 1.8 billion synaptic events per second, with the most complex simulations consuming less than 1 Watt per SpiNNaker chip.","PeriodicalId":376975,"journal":{"name":"The 2013 International Joint Conference on Neural Networks (IJCNN)","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126401781","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 : 2013-08-04DOI: 10.1109/IJCNN.2013.6706781
Vaibhav Gandhi, T. Mcginnity
A filtering methodology inspired by the principles of quantum mechanics and incorporating the well-known Schrodinger wave equation is investigated for the first time for filtering EMG signals. This architecture, referred to as a Recurrent Quantum Neural Network (RQNN) can characterize a non-stationary stochastic signal as time varying wave packets. An unsupervised learning rule allows the RQNN to capture the statistical behaviour of the input signal and facilitates estimation of an EMG signal embedded in noise with unknown characteristics. Results from a number of benchmark tests show that simple signals such as DC, staircase DC and sinusoidal signals embedded with a high level of noise can be accurately filtered. Particle swarm optimization is employed to select RQNN model parameters for filtering simple signals. In this paper, we present the RQNN filtering procedure, using heuristically selected parameters, to be applied to a new thirteen class EMG based finger movement detection system, for emulation in a Shadow Robotics robot hand. It is shown that the RQNN EMG filtering improves the classification performance compared to using only the raw EMG signals, across multiple feature extraction approaches and subjects. Effective control of the robot hand is demonstrated.
{"title":"Quantum neural network based surface EMG signal filtering for control of robotic hand","authors":"Vaibhav Gandhi, T. Mcginnity","doi":"10.1109/IJCNN.2013.6706781","DOIUrl":"https://doi.org/10.1109/IJCNN.2013.6706781","url":null,"abstract":"A filtering methodology inspired by the principles of quantum mechanics and incorporating the well-known Schrodinger wave equation is investigated for the first time for filtering EMG signals. This architecture, referred to as a Recurrent Quantum Neural Network (RQNN) can characterize a non-stationary stochastic signal as time varying wave packets. An unsupervised learning rule allows the RQNN to capture the statistical behaviour of the input signal and facilitates estimation of an EMG signal embedded in noise with unknown characteristics. Results from a number of benchmark tests show that simple signals such as DC, staircase DC and sinusoidal signals embedded with a high level of noise can be accurately filtered. Particle swarm optimization is employed to select RQNN model parameters for filtering simple signals. In this paper, we present the RQNN filtering procedure, using heuristically selected parameters, to be applied to a new thirteen class EMG based finger movement detection system, for emulation in a Shadow Robotics robot hand. It is shown that the RQNN EMG filtering improves the classification performance compared to using only the raw EMG signals, across multiple feature extraction approaches and subjects. Effective control of the robot hand is demonstrated.","PeriodicalId":376975,"journal":{"name":"The 2013 International Joint Conference on Neural Networks (IJCNN)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134592368","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}
When two or more sound detectors are available, interaural time differences may be used to determine the direction of a sound's origin. This process, known as sound localization, is performed in mammals via the auditory pathways of the head and by computation in the brain. The Jeffress Model successfully describes the mechanism by exploiting coincidence detector neurons in conjunction with delay lines. However, one of the difficulties of using this model on neural simulators is that it requires timing accuracies which are much finer than the typical 1 ms resolution provided by simulation platforms. One solution is clearly to reduce the simulation's time step, but in this paper we also explore the use of population coding to represent more precise timing information without changing the simulation's timing resolution. The implementation of both the Jeffress and population coded models are contrasted, together with their results, which show that population coding is indeed able to provide successful sound localization.
{"title":"Modeling populations of spiking neurons for fine timing sound localization","authors":"Qian Liu, Cameron Patterson, S. Furber, Zhangqin Huang, Yibin Hou, Huibing Zhang","doi":"10.1109/IJCNN.2013.6706931","DOIUrl":"https://doi.org/10.1109/IJCNN.2013.6706931","url":null,"abstract":"When two or more sound detectors are available, interaural time differences may be used to determine the direction of a sound's origin. This process, known as sound localization, is performed in mammals via the auditory pathways of the head and by computation in the brain. The Jeffress Model successfully describes the mechanism by exploiting coincidence detector neurons in conjunction with delay lines. However, one of the difficulties of using this model on neural simulators is that it requires timing accuracies which are much finer than the typical 1 ms resolution provided by simulation platforms. One solution is clearly to reduce the simulation's time step, but in this paper we also explore the use of population coding to represent more precise timing information without changing the simulation's timing resolution. The implementation of both the Jeffress and population coded models are contrasted, together with their results, which show that population coding is indeed able to provide successful sound localization.","PeriodicalId":376975,"journal":{"name":"The 2013 International Joint Conference on Neural Networks (IJCNN)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132362864","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 : 2013-08-04DOI: 10.1109/IJCNN.2013.6706887
Alexander D. Rast, J. Partzsch, C. Mayr, J. Schemmel, Stefan Hartmann, L. Plana, S. Temple, D. Lester, R. Schüffny, S. Furber
With neuromorphic hardware rapidly moving towards large-scale, possibly immovable systems capable of implementing brain-scale neural models in hardware, there is an emerging need to be able to integrate multi-system combinations of sensors and cortical processors over distributed, multisite configurations. If there were a standard, direct interface allowing large systems to communicate using native signalling, it would be possible to use heterogeneous resources efficiently according to their task suitability. We propose a UDP-based AER spiking interface that permits direct bidirectional spike communications over standard networks, and demonstrate a practical implementation with two large-scale neuromorphic systems, BrainScaleS and SpiNNaker. Internally, the interfaces at either end appear as interceptors which decode and encode spikes in a standardised AER address format onto UDP frames. The system is able to run a spiking neural network distributed over the two systems, in both a side-by-side setup with a direct cable link and over the Internet between 2 widely spaced sites. Such a model not only realises a solution for connecting remote sensors or processors to a large, central neuromorphic simulation platform, but also opens possibilities for interesting automated remote neural control, such as parameter tuning, for large, complex neural systems, and suggests methods to overcome differences in timescale and simulation model between different platforms. With its entirely standard protocol and physical layer, the interface makes large neuromorphic systems a distributed, accessible resource available to all.
{"title":"A location-independent direct link neuromorphic interface","authors":"Alexander D. Rast, J. Partzsch, C. Mayr, J. Schemmel, Stefan Hartmann, L. Plana, S. Temple, D. Lester, R. Schüffny, S. Furber","doi":"10.1109/IJCNN.2013.6706887","DOIUrl":"https://doi.org/10.1109/IJCNN.2013.6706887","url":null,"abstract":"With neuromorphic hardware rapidly moving towards large-scale, possibly immovable systems capable of implementing brain-scale neural models in hardware, there is an emerging need to be able to integrate multi-system combinations of sensors and cortical processors over distributed, multisite configurations. If there were a standard, direct interface allowing large systems to communicate using native signalling, it would be possible to use heterogeneous resources efficiently according to their task suitability. We propose a UDP-based AER spiking interface that permits direct bidirectional spike communications over standard networks, and demonstrate a practical implementation with two large-scale neuromorphic systems, BrainScaleS and SpiNNaker. Internally, the interfaces at either end appear as interceptors which decode and encode spikes in a standardised AER address format onto UDP frames. The system is able to run a spiking neural network distributed over the two systems, in both a side-by-side setup with a direct cable link and over the Internet between 2 widely spaced sites. Such a model not only realises a solution for connecting remote sensors or processors to a large, central neuromorphic simulation platform, but also opens possibilities for interesting automated remote neural control, such as parameter tuning, for large, complex neural systems, and suggests methods to overcome differences in timescale and simulation model between different platforms. With its entirely standard protocol and physical layer, the interface makes large neuromorphic systems a distributed, accessible resource available to all.","PeriodicalId":376975,"journal":{"name":"The 2013 International Joint Conference on Neural Networks (IJCNN)","volume":"67 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121539314","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 : 2013-08-04DOI: 10.1109/IJCNN.2013.6707044
Jean-Charles Lamirel
This paper deals with a new feature selection and feature contrasting approach for classification of highly imbalanced textual data with a high degree of similarity between associated classes. An example of such classification context is illustrated by the task of classifying bibliographic references into a patent classification scheme. This task represents one of the domains of investigation of the QUAERO project, with the final goal of helping experts to evaluate upcoming patents through the use of related research.
{"title":"Dealing with highly imbalanced textual data gathered into similar classes","authors":"Jean-Charles Lamirel","doi":"10.1109/IJCNN.2013.6707044","DOIUrl":"https://doi.org/10.1109/IJCNN.2013.6707044","url":null,"abstract":"This paper deals with a new feature selection and feature contrasting approach for classification of highly imbalanced textual data with a high degree of similarity between associated classes. An example of such classification context is illustrated by the task of classifying bibliographic references into a patent classification scheme. This task represents one of the domains of investigation of the QUAERO project, with the final goal of helping experts to evaluate upcoming patents through the use of related research.","PeriodicalId":376975,"journal":{"name":"The 2013 International Joint Conference on Neural Networks (IJCNN)","volume":"95 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124226950","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 : 2013-08-04DOI: 10.1109/IJCNN.2013.6707121
Thanh-Nghi Doan, Thanh-Nghi Do, F. Poulet
ImageNet dataset [1] with more than 14M images and 21K classes makes the problem of visual classification more difficult to deal with. One of the most difficult tasks is to train a fast and accurate classifier on computers with limited memory resource. In this paper, we address this challenge by extending the state-of-the-art large scale classifier Power Mean SVM (PmSVM) proposed by Jianxin Wu [2] in three ways: (1) An incremental learning for PmSVM, (2) A balanced bagging algorithm for training binary classifiers, (3) Parallelize the training process of classifiers with several multi-core computers. Our approach is evaluated on 1K classes of ImageNet (ILSVRC 1000 [3]). The evaluation shows that our approach can save up to 84.34% memory usage and the training process is 297 times faster than the original implementation and 1508 times faster than the state-of-the-art linear classifier (LIBLINEAR [4]).
ImageNet数据集[1]拥有超过14M张图像和21K个类，使得视觉分类问题更加难以处理。在内存有限的计算机上训练快速准确的分类器是最困难的任务之一。在本文中，我们通过三种方式扩展了由Jianxin Wu[2]提出的最先进的大规模分类器Power Mean SVM (PmSVM)来解决这一挑战:(1)PmSVM的增量学习，(2)训练二分类器的平衡bagging算法，(3)在多核计算机上并行化分类器的训练过程。我们的方法在1K个ImageNet类(ILSVRC 1000[3])上进行了评估。评估表明，我们的方法可以节省高达84.34%的内存使用，训练过程比原始实现快297倍，比最先进的线性分类器(LIBLINEAR[4])快1508倍。
{"title":"Parallel incremental SVM for classifying million images with very high-dimensional signatures into thousand classes","authors":"Thanh-Nghi Doan, Thanh-Nghi Do, F. Poulet","doi":"10.1109/IJCNN.2013.6707121","DOIUrl":"https://doi.org/10.1109/IJCNN.2013.6707121","url":null,"abstract":"ImageNet dataset [1] with more than 14M images and 21K classes makes the problem of visual classification more difficult to deal with. One of the most difficult tasks is to train a fast and accurate classifier on computers with limited memory resource. In this paper, we address this challenge by extending the state-of-the-art large scale classifier Power Mean SVM (PmSVM) proposed by Jianxin Wu [2] in three ways: (1) An incremental learning for PmSVM, (2) A balanced bagging algorithm for training binary classifiers, (3) Parallelize the training process of classifiers with several multi-core computers. Our approach is evaluated on 1K classes of ImageNet (ILSVRC 1000 [3]). The evaluation shows that our approach can save up to 84.34% memory usage and the training process is 297 times faster than the original implementation and 1508 times faster than the state-of-the-art linear classifier (LIBLINEAR [4]).","PeriodicalId":376975,"journal":{"name":"The 2013 International Joint Conference on Neural Networks (IJCNN)","volume":"61 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126294813","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：