A deep learning approach to multiple kernel fusion

Kernel fusion is a popular and effective approach for combining multiple features that characterize different aspects of data. Traditional approaches for Multiple Kernel Learning (MKL) attempt to learn the parameters for combining the kernels through sophisticated optimization procedures. In this paper, we propose an alternative approach that creates dense embeddings for data using the kernel similarities and adopts a deep neural network architecture for fusing the embeddings. In order to improve the effectiveness of this network, we introduce the kernel dropout regularization strategy coupled with the use of an expanded set of composition kernels. Experiment results on a real-world activity recognition dataset show that the proposed architecture is effective in fusing kernels and achieves state-of-the-art performance.

[1]  Cheng Soon Ong,et al.  Multiclass multiple kernel learning , 2007, ICML '07.

[2]  S. V. N. Vishwanathan,et al.  SPF-GMKL: generalized multiple kernel learning with a million kernels , 2012, KDD.

[3]  Shyam Visweswaran,et al.  Deep Multiple Kernel Learning , 2013, 2013 12th International Conference on Machine Learning and Applications.

[4]  Karthikeyan Natesan Ramamurthy,et al.  Learning Stable Multilevel Dictionaries for Sparse Representations , 2013, IEEE Transactions on Neural Networks and Learning Systems.

[5]  Mi Zhang,et al.  Human Daily Activity Recognition With Sparse Representation Using Wearable Sensors , 2013, IEEE Journal of Biomedical and Health Informatics.

[6]  S. V. N. Vishwanathan,et al.  Multiple Kernel Learning and the SMO Algorithm , 2010, NIPS.

[7]  Karthikeyan Natesan Ramamurthy,et al.  Auto-context modeling using multiple Kernel learning , 2016, 2016 IEEE International Conference on Image Processing (ICIP).

[8]  Pavan K. Turaga,et al.  Shape Distributions of Nonlinear Dynamical Systems for Video-Based Inference , 2016, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[9]  Jeffrey Dean,et al.  Distributed Representations of Words and Phrases and their Compositionality , 2013, NIPS.

[10]  Raghvendra Kannao,et al.  TV News Commercials Detection using Success based Locally Weighted Kernel Combination , 2015, ArXiv.

[11]  Gerlof Bouma,et al.  Normalized (pointwise) mutual information in collocation extraction , 2009 .

[12]  Lawrence K. Saul,et al.  Kernel Methods for Deep Learning , 2009, NIPS.

[13]  Karthikeyan Natesan Ramamurthy,et al.  Multiple Kernel Sparse Representations for Supervised and Unsupervised Learning , 2013, IEEE Transactions on Image Processing.

[14]  Shuicheng Yan,et al.  Graph embedding: a general framework for dimensionality reduction , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05).

[15]  Sergey Ioffe,et al.  Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift , 2015, ICML.

[16]  Ethem Alpaydin,et al.  Localized multiple kernel learning , 2008, ICML '08.

[17]  Juhan Nam,et al.  Multimodal Deep Learning , 2011, ICML.

[18]  Subhransu Maji,et al.  Classification using intersection kernel support vector machines is efficient , 2008, 2008 IEEE Conference on Computer Vision and Pattern Recognition.

[19]  Qinghua Hu,et al.  Heterogeneous Features Integration via Semi-supervised Multi-modal Deep Networks , 2015, ICONIP.

[20]  Karthikeyan Natesan Ramamurthy,et al.  Consensus inference on mobile phone sensors for activity recognition , 2016, 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[21]  Raghvendra Kannao,et al.  TV Commercial Detection Using Success Based Locally Weighted Kernel Combination , 2016, MMM.

[22]  Heikki Mannila,et al.  Random projection in dimensionality reduction: applications to image and text data , 2001, KDD '01.

[23]  Jason Weston,et al.  A kernel method for multi-labelled classification , 2001, NIPS.

[24]  Karthikeyan Natesan Ramamurthy,et al.  Kernel Sparse Models for Automated Tumor Segmentation , 2013, Int. J. Artif. Intell. Tools.

[25]  Ahmed M. Elgammal,et al.  Inferring 3D body pose from silhouettes using activity manifold learning , 2004, Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2004. CVPR 2004..

[26]  Mi Zhang,et al.  USC-HAD: a daily activity dataset for ubiquitous activity recognition using wearable sensors , 2012, UbiComp.

[27]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[28]  Shie Mannor,et al.  Activity and Gait Recognition with Time-Delay Embeddings , 2010, AAAI.

[29]  Omer Levy,et al.  Neural Word Embedding as Implicit Matrix Factorization , 2014, NIPS.