Header

UZH-Logo

Maintenance Infos

Delta networks for optimized recurrent network computation


Neil, D; Lee, J-H; Delbruck, T; Liu, S-C (2017). Delta networks for optimized recurrent network computation. In: 34th International Conference on Machine Learning 2017, Sydney, 6 August 2017 - 11 August 2017.

Abstract

Many neural networks exhibit stability in their activation patterns over time in response to inputs from sensors operating under real-world conditions. By capitalizing on this property of natural signals, we propose a Recurrent Neural Network (RNN) architecture called a delta network in which each neuron transmits its value only when the change in its activation exceeds a threshold. The execution of RNNs as delta networks is attractive because their states must be stored and fetched at every timestep, unlike in convolutional neural networks (CNNs). We show that a naive run-time delta network implementation offers modest improvements on the number of memory accesses and computes, but optimized training techniques confer higher accuracy at higher speedup. With these optimizations, we demonstrate a 9X reduction in cost with negligible loss of accuracy for the TIDIGITS audio digit recognition benchmark. Similarly, on the large Wall Street Journal (WSJ) speech recognition benchmark, pretrained networks can also be greatly accelerated as delta networks and trained delta networks show a 5.7X improvement with negligible loss of accuracy. Finally, on an endto-end CNN-RNN network trained for steering angle prediction in a driving dataset, the RNN cost can be reduced by a substantial 100X.

Abstract

Many neural networks exhibit stability in their activation patterns over time in response to inputs from sensors operating under real-world conditions. By capitalizing on this property of natural signals, we propose a Recurrent Neural Network (RNN) architecture called a delta network in which each neuron transmits its value only when the change in its activation exceeds a threshold. The execution of RNNs as delta networks is attractive because their states must be stored and fetched at every timestep, unlike in convolutional neural networks (CNNs). We show that a naive run-time delta network implementation offers modest improvements on the number of memory accesses and computes, but optimized training techniques confer higher accuracy at higher speedup. With these optimizations, we demonstrate a 9X reduction in cost with negligible loss of accuracy for the TIDIGITS audio digit recognition benchmark. Similarly, on the large Wall Street Journal (WSJ) speech recognition benchmark, pretrained networks can also be greatly accelerated as delta networks and trained delta networks show a 5.7X improvement with negligible loss of accuracy. Finally, on an endto-end CNN-RNN network trained for steering angle prediction in a driving dataset, the RNN cost can be reduced by a substantial 100X.

Statistics

Downloads

11 downloads since deposited on 23 Feb 2018
11 downloads since 12 months
Detailed statistics

Additional indexing

Item Type:Conference or Workshop Item (Paper), refereed, original work
Communities & Collections:07 Faculty of Science > Institute of Neuroinformatics
Dewey Decimal Classification:570 Life sciences; biology
Language:English
Event End Date:11 August 2017
Deposited On:23 Feb 2018 09:33
Last Modified:31 Jul 2018 05:12
Publisher:Proceedings of the 34th International Conference on Machine Learning
Series Name:Proceedings of the 34th International Conference on Machine Learning
OA Status:Green
Free access at:Official URL. An embargo period may apply.
Official URL:http://proceedings.mlr.press/v70/neil17a/neil17a.pdf

Download

Download PDF  'Delta networks for optimized recurrent network computation'.
Preview
Content: Published Version
Filetype: PDF
Size: 877kB