Learning Channel-Wise Ordered Aggregations in Deep Neural Networks

One of the most common techniques for approaching image classification problems are Deep Neural Networks. These systems are capable of classifying images with different levels of detail at different levels of detail, with an accuracy that sometimes can surpass even manual classification by humans. Most common architectures for Deep Neural Networks are based on convolutional layers, which perform at the same time a convolution on each input channel and a linear aggregation on the convoluted channels. In this work, we develop a new method for augmenting the information of a layer inside a Deep Neural Network using channel-wise ordered aggregations. We develop a new layer that can be placed at different points inside a Deep Neural Network. This layer takes the feature maps of the previous layer and adds new feature maps by applying several channel-wise ordered aggregations based on learned weighting vectors. We perform several experiments introducing this layer in a VGG neural network and study the impact of the new layer, obtaining better accuracy scores over a sample dataset based on ImageNet. We also study the convergence and evolution of the weighting vectors of the new layers over the learning process, which gives a better understanding of the way the system is exploiting the additional information to gain new knowledge.

[1]  H. B. Mann,et al.  On a Test of Whether one of Two Random Variables is Stochastically Larger than the Other , 1947 .

[2]  Leslie N. Smith,et al.  A disciplined approach to neural network hyper-parameters: Part 1 - learning rate, batch size, momentum, and weight decay , 2018, ArXiv.

[3]  Curt H. Davis,et al.  Fusion of Deep Convolutional Neural Networks for Land Cover Classification of High-Resolution Imagery , 2017, IEEE Geoscience and Remote Sensing Letters.

[4]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[5]  Li Fei-Fei,et al.  ImageNet: A large-scale hierarchical image database , 2009, CVPR.

[6]  Derek T. Anderson,et al.  Introducing Fuzzy Layers for Deep Learning , 2019, 2019 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE).

[7]  Xiaoxiao Du,et al.  Multiple Instance Choquet Integral Classifier Fusion and Regression for Remote Sensing Applications , 2018, IEEE Transactions on Geoscience and Remote Sensing.

[8]  Ronald R. Yager,et al.  On ordered weighted averaging aggregation operators in multicriteria decisionmaking , 1988, IEEE Trans. Syst. Man Cybern..

[9]  Curt H. Davis,et al.  Enhanced Fusion of Deep Neural Networks for Classification of Benchmark High-Resolution Image Data Sets , 2018, IEEE Geoscience and Remote Sensing Letters.

[10]  Humberto Bustince,et al.  Using the Choquet Integral in the Pooling Layer in Deep Learning Networks , 2018, NAFIPS.

[11]  Derek T. Anderson,et al.  Fuzzy Choquet Integration of Deep Convolutional Neural Networks for Remote Sensing , 2018 .

[12]  James M. Keller,et al.  Linear Order Statistic Neuron , 2019, 2019 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE).

[13]  L. Rudin,et al.  Nonlinear total variation based noise removal algorithms , 1992 .

[14]  Humberto Bustince,et al.  Simulating the Behaviour of Choquet-Like (pre) Aggregation Functions for Image Resizing in the Pooling Layer of Deep Learning Networks , 2019, IFSA/NAFIPS.