Lazy Evaluation of Convolutional Filters

05/27/2016

∙

In this paper we propose a technique which avoids the evaluation of certain convolutional filters in a deep neural network. This allows to trade-off the accuracy of a deep neural network with the computational and memory requirements. This is especially important on a constrained device unable to hold all the weights of the network in memory.

READ FULL TEXT

Lazy Evaluation of Convolutional Filters

Binary Neural Network for Speaker Verification

An Evaluation of Software Sketches

RePr: Improved Training of Convolutional Filters

Enhancing the Performance of Convolutional Neural Networks on Quality Degraded Datasets

Low-Rank Training of Deep Neural Networks for Emerging Memory Technology

An Enhanced SCMA Detector Enabled by Deep Neural Network

MimosaNet: An Unrobust Neural Network Preventing Model Stealing

Lazy Evaluation of Convolutional Filters

Related Research

Binary Neural Network for Speaker Verification

An Evaluation of Software Sketches

RePr: Improved Training of Convolutional Filters

Enhancing the Performance of Convolutional Neural Networks on Quality Degraded Datasets

Low-Rank Training of Deep Neural Networks for Emerging Memory Technology

An Enhanced SCMA Detector Enabled by Deep Neural Network

MimosaNet: An Unrobust Neural Network Preventing Model Stealing