Deep learning and the information bottleneck principle
Naftali Tishby,Noga Zaslavsky +1 more
- pp 1-5
TLDR
It is argued that both the optimal architecture, number of layers and features/connections at each layer, are related to the bifurcation points of the information bottleneck tradeoff, namely, relevant compression of the input layer with respect to the output layer.Abstract:
Deep Neural Networks (DNNs) are analyzed via the theoretical framework of the information bottleneck (IB) principle. We first show that any DNN can be quantified by the mutual information between the layers and the input and output variables. Using this representation we can calculate the optimal information theoretic limits of the DNN and obtain finite sample generalization bounds. The advantage of getting closer to the theoretical limit is quantifiable both by the generalization bound and by the network's simplicity. We argue that both the optimal architecture, number of layers and features/connections at each layer, are related to the bifurcation points of the information bottleneck tradeoff, namely, relevant compression of the input layer with respect to the output layer. The hierarchical representations at the layered network naturally correspond to the structural phase transitions along the information curve. We believe that this new insight can lead to new optimality bounds and deep learning algorithms.read more
Citations
More filters
Journal ArticleDOI
Optimal transfer protocol by incremental layer defrosting
TL;DR: In this article , the authors show that the simple transfer learning protocol is often sub-optimal and the largest performance gain may be achieved when smaller portions of the pre-trained network are kept frozen, which depends on the amount of available training data and the degree of source-target task correlation.
Renormalization Group-Motivated Learning
TL;DR: In this paper , an information-bottleneck-like trade-off is proposed to compress the data to maximize the correlation between the two data points to be compressed while minimizing the correlations between the paired data and other data points.
Proceedings ArticleDOI
A Novel Structure of Convolutional Layers with a Higher Performance-Complexity Ratio for Semantic Segmentation
Yalong Jiang,Zheru Chi +1 more
TL;DR: Experimental results on the segmentation of the PASCAL Person Parts Dataset show that the linear dependency among convolutional kernels is an important factor determining the capacity of a CNN model, and a measure based on the mutual information between hidden activations and inputs/outputs to compute the capacity is proposed.
Journal ArticleDOI
T-RECX: Tiny-Resource Efficient Convolutional Neural Networks with Early-Exit
TL;DR: The challenges of adding early-exits to state-of-the-art tiny-CNNs are discussed and an early-exit architecture, T-R EC X, is devised that addresses these challenges and develops a method to alleviate the effect of network overthinking at the exit by leveraging the high-level representations learned by the early- exit.
Posted ContentDOI
Modeling transcriptional profiles of gene perturbation with deep neural network
TL;DR: In this article, the problem of inferring gene perturbation based on a reference database was reframed into a classification task and evaluated the application of deep neural network models to address this problem.
References
More filters
Proceedings Article
ImageNet Classification with Deep Convolutional Neural Networks
TL;DR: The state-of-the-art performance of CNNs was achieved by Deep Convolutional Neural Networks (DCNNs) as discussed by the authors, which consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully-connected layers with a final 1000-way softmax.
Book
Elements of information theory
Thomas M. Cover,Joy A. Thomas +1 more
TL;DR: The author examines the role of entropy, inequality, and randomness in the design of codes and the construction of codes in the rapidly changing environment.
Journal ArticleDOI
Reducing the Dimensionality of Data with Neural Networks
TL;DR: In this article, an effective way of initializing the weights that allows deep autoencoder networks to learn low-dimensional codes that work much better than principal components analysis as a tool to reduce the dimensionality of data is described.
Journal ArticleDOI
Representation Learning: A Review and New Perspectives
TL;DR: Recent work in the area of unsupervised feature learning and deep learning is reviewed, covering advances in probabilistic models, autoencoders, manifold learning, and deep networks.
Book
Learning Deep Architectures for AI
TL;DR: The motivations and principles regarding learning algorithms for deep architectures, in particular those exploiting as building blocks unsupervised learning of single-layer modelssuch as Restricted Boltzmann Machines, used to construct deeper models such as Deep Belief Networks are discussed.