论文信息 - Deep Learning Vector Quantization

Deep Learning Vector Quantization

While deep neural nets (DNN’s) achieve impressive performance on image recognition tasks, previous studies have reported that DNN’s give high confidence predictions for unrecognizable images. Motivated by the observation that such fooling examples might be caused by the extrapolating nature of the log-softmax, we propose to combine neural networks with Learning Vector Quantization (LVQ). Our proposed method, called Deep LVQ (DLVQ), achieves comparable performance on MNIST while being more robust against fooling and adversarial examples.

Aaron C. Courville | Roland Memisevic | Harm de Vries

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

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

[3] Jonathon Shlens,et al. Explaining and Harnessing Adversarial Examples , 2014, ICLR.

[4] Thomas Villmann,et al. Supervised Neural Gas with General Similarity Measure , 2005, Neural Processing Letters.

[5] Joan Bruna,et al. Intriguing properties of neural networks , 2013, ICLR.

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

[7] Geoffrey E. Hinton,et al. ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[8] Pascal Vincent,et al. Unsupervised Feature Learning and Deep Learning: A Review and New Perspectives , 2012, ArXiv.

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

[10] Jason Yosinski,et al. Deep neural networks are easily fooled: High confidence predictions for unrecognizable images , 2014, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).