Improving the Neural GPU Architecture for Algorithm Learning

Karlis Freivalds, Renars Liepins
2/28/2017
cs.NE

Official code: LUMII-Syslab/DNGPU (13 stars, MIT, last commit 2018-07-04).

Abstract

Algorithm learning is a core problem in artificial intelligence with significant implications on automation level that can be achieved by machines. Recently deep learning methods are emerging for synthesizing an algorithm from its input-output examples, the most successful being the Neural GPU, capable of learning multiplication. We present several improvements to the Neural GPU that substantially reduces training time and improves generalization. We introduce a new technique - hard nonlinearities with saturation costs- that has general applicability. We also introduce a technique of diagonal gates that can be applied to active-memory models. The proposed architecture is the first capable of learning decimal multiplication end-to-end.

View on arXivView PDF

Code Implementations(1)

LUMII-Syslab/DNGPUOfficial100%
MIT

Code for reproducing key results in the paper "Improving the Neural GPU Architecture for Algorithm Learning" by Karlis Freivalds, Renars Liepins.

134PythonFeb 27, 20178 years agoMIT

Similar Papers

Is Integer Arithmetic Enough for Deep Learning Training?

84% match2022

MCTensor: A High-Precision Deep Learning Library with Multi-Component Floating-Point

83% match1 repo82022

RLlib: Abstractions for Distributed Reinforcement Learning

82% match1 repo43.6k2017

Ristretto: Hardware-Oriented Approximation of Convolutional Neural Networks

82% match1 repo2.2k2016

Exploring a Cognitive Architecture for Learning Arithmetic Equations

82% match2024

Continual Learning with Deep Learning Methods in an Application-Oriented Context

82% match2022

Cite this paper

@article{freivalds2017improving,
  title  = {Improving the Neural GPU Architecture for Algorithm Learning},
  author = {Karlis Freivalds and Renars Liepins},
  year   = {2017},
  eprint = {1702.08727},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1702.08727}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1702.08727.svg)](https://codeofpaper.com/paper/1702.08727)

Discussion