Combining offline and online classifiers for life-long learning

One of the greatest challenges of life-long learning architectures is how to efficiently and reliably cope with the stability-plasticity dilemma. We propose an extension of a flexible system combining a static offline classifier and an incremental online classifier that is well suited for life-long learning scenarios. The pre-trained offline classifier preserves ground knowledge that should be respected during training, while the online classifier enables learning of new or specific information encountered during use. The combination is realised by a dynamic classifier selection strategy based on confidences of both ingredients. We report exemplary results of this architecture for the case of learning vector quantization (LVQ) for several data sets, thereby including an extensive comparison to alternative state of the art algorithms for incremental learning such as incremental generalised LVQ and the support vector machine.

[1]  Horst-Michael Groß,et al.  A vision architecture for unconstrained and incremental learning of multiple categories , 2009, Memetic Comput..

[2]  Heiko Wersing,et al.  Efficient rejection strategies for prototype-based classification , 2015, Neurocomputing.

[3]  Barbara Hammer,et al.  Patch clustering for massive data sets , 2009, Neurocomputing.

[4]  Robi Polikar,et al.  COMPOSE: A Semisupervised Learning Framework for Initially Labeled Nonstationary Streaming Data , 2014, IEEE Transactions on Neural Networks and Learning Systems.

[5]  Marc Sebban,et al.  A Survey on Metric Learning for Feature Vectors and Structured Data , 2013, ArXiv.

[6]  B.V. Dasarathy,et al.  A composite classifier system design: Concepts and methodology , 1979, Proceedings of the IEEE.

[7]  Kevin W. Bowyer,et al.  Combination of Multiple Classifiers Using Local Accuracy Estimates , 1997, IEEE Trans. Pattern Anal. Mach. Intell..

[8]  Slobodan Vucetic,et al.  Learning Vector Quantization with adaptive prototype addition and removal , 2009, 2009 International Joint Conference on Neural Networks.

[9]  Katsumi Inoue,et al.  Learning revised models for planning in adaptive systems , 2013, 2013 35th International Conference on Software Engineering (ICSE).

[10]  Heiko Wersing,et al.  Certainty-based prototype insertion/deletion for classification with metric adaptation , 2015, ESANN.

[11]  Vladimir Vapnik,et al.  An overview of statistical learning theory , 1999, IEEE Trans. Neural Networks.

[12]  Nicolai Petkov,et al.  Adaptive Matrices and Filters for Color Texture Classification , 2012, Journal of Mathematical Imaging and Vision.

[13]  Heiko Wersing,et al.  A biologically motivated visual memory architecture for online learning of objects , 2008, Neural Networks.

[14]  Ye Xu,et al.  An incremental learning vector quantization algorithm for pattern classification , 2010, Neural Computing and Applications.

[15]  Pablo A. Estévez,et al.  A review of learning vector quantization classifiers , 2013, Neural Computing and Applications.

[16]  Robi Polikar,et al.  Guest Editorial Learning in Nonstationary and Evolving Environments , 2014, IEEE Transactions on Neural Networks and Learning Systems.

[17]  Klaus Obermayer,et al.  Soft Learning Vector Quantization , 2003, Neural Computation.

[18]  Xin Yao,et al.  DDD: A New Ensemble Approach for Dealing with Concept Drift , 2012, IEEE Transactions on Knowledge and Data Engineering.

[19]  Marcus A. Maloof,et al.  Dynamic Weighted Majority: An Ensemble Method for Drifting Concepts , 2007, J. Mach. Learn. Res..

[20]  Michael Biehl,et al.  Analysis of Flow Cytometry Data by Matrix Relevance Learning Vector Quantization , 2013, PloS one.

[21]  Horst-Michael Groß,et al.  A life-long learning vector quantization approach for interactive learning of multiple categories , 2012, Neural Networks.

[22]  Amar Mitiche,et al.  Classifier combination for hand-printed digit recognition , 1993, Proceedings of 2nd International Conference on Document Analysis and Recognition (ICDAR '93).

[23]  Heiko Wersing,et al.  Rapid Online Learning of Objects in a Biologically Motivated Recognition Architecture , 2005, DAGM-Symposium.

[24]  Michael Biehl,et al.  Adaptive Relevance Matrices in Learning Vector Quantization , 2009, Neural Computation.

[25]  Gert Cauwenberghs,et al.  Incremental and Decremental Support Vector Machine Learning , 2000, NIPS.

[26]  Gert Cauwenberghs,et al.  SVM incremental learning, adaptation and optimization , 2003, Proceedings of the International Joint Conference on Neural Networks, 2003..

[27]  Jeffrey Queißer,et al.  Using context for the combination of off-line and on-line learning , 2012 .

[28]  Bernhard Sendhoff,et al.  Alleviating Catastrophic Forgetting via Multi-Objective Learning , 2006, The 2006 IEEE International Joint Conference on Neural Network Proceedings.

[29]  Teuvo Kohonen,et al.  Self-organization and associative memory: 3rd edition , 1989 .

[30]  Haibo He,et al.  Incremental Learning From Stream Data , 2011, IEEE Transactions on Neural Networks.

[31]  Hamid Beigy,et al.  Using a classifier pool in accuracy based tracking of recurring concepts in data stream classification , 2013, Evol. Syst..

[32]  Sebastian Thrun,et al.  Online Speed Adaptation Using Supervised Learning for High-Speed, Off-Road Autonomous Driving , 2007, IJCAI.

[33]  Yoshua Bengio,et al.  An Empirical Investigation of Catastrophic Forgeting in Gradient-Based Neural Networks , 2013, ICLR.

[34]  Martin A. Riedmiller,et al.  Incremental GRLVQ: Learning relevant features for 3D object recognition , 2008, Neurocomputing.

[35]  Vasant Honavar,et al.  Learn++: an incremental learning algorithm for supervised neural networks , 2001, IEEE Trans. Syst. Man Cybern. Part C.

[36]  Sebastian Thrun,et al.  Lifelong robot learning , 1993, Robotics Auton. Syst..

[37]  Thomas Villmann,et al.  Distance Measures for Prototype Based Classification , 2013, BrainComp.

[38]  Michael Biehl,et al.  Insightful stress detection from physiology modalities using Learning Vector Quantization , 2015, Neurocomputing.

[39]  Frank-Michael Schleif,et al.  Adaptive conformal semi-supervised vector quantization for dissimilarity data , 2014, Pattern Recognit. Lett..

[40]  Atsushi Sato,et al.  Generalized Learning Vector Quantization , 1995, NIPS.

[41]  João Gama,et al.  A survey on concept drift adaptation , 2014, ACM Comput. Surv..

[42]  Gyan Bhanot,et al.  Inter-species prediction of protein phosphorylation in the sbv IMPROVER species translation challenge , 2015, Bioinform..

[43]  Amaury Habrard,et al.  Robustness and generalization for metric learning , 2012, Neurocomputing.

[44]  Laurens van der Maaten.  Matlab Toolbox for Dimensionality Reduction Laurens van der Maaten , 2007 .

[45]  Qiang Yang,et al.  Lifelong Machine Learning Systems: Beyond Learning Algorithms , 2013, AAAI Spring Symposium: Lifelong Machine Learning.

[46]  Barbara Hammer,et al.  Parametric nonlinear dimensionality reduction using kernel t-SNE , 2015, Neurocomputing.

[47]  Thomas Villmann,et al.  Limited Rank Matrix Learning, discriminative dimension reduction and visualization , 2012, Neural Networks.

[48]  Ye Xu,et al.  An Online Incremental Learning Vector Quantization , 2009, PAKDD.

[49]  Luiz Eduardo Soares de Oliveira,et al.  Dynamic selection of classifiers - A comprehensive review , 2014, Pattern Recognit..

[50]  Koby Crammer,et al.  Adaptive regularization of weight vectors , 2009, Machine Learning.