Learning and recognition of multiple fluctuating temporal patterns using S-CTRNN

Shingo Murata, Hiroaki Arie, Tetsuya Ogata, Jun Tani, Shigeki Sugano

Research output: Chapter in Book/Report/Conference proceedingConference contribution

8 Citations (Scopus)

Abstract

In the present study, we demonstrate the learning and recognition capabilities of our recently proposed recurrent neural network (RNN) model called stochastic continuous-time RNN (S-CTRNN). S-CTRNN can learn to predict not only the mean but also the variance of the next state of the learning targets. The network parameters consisting of weights, biases, and initial states of context neurons are optimized through maximum likelihood estimation (MLE) using the gradient descent method. First, we clarify the essential difference between the learning capabilities of conventional CTRNN and S-CTRNN by analyzing the results of a numerical experiment in which multiple fluctuating temporal patterns were used as training data, where the variance of the Gaussian noise varied among the patterns. Furthermore, we also show that the trained S-CTRNN can recognize given fluctuating patterns by inferring the initial states that can reproduce the patterns through the same MLE scheme as that used for network training.

Original languageEnglish
Title of host publicationArtificial Neural Networks and Machine Learning, ICANN 2014 - 24th International Conference on Artificial Neural Networks, Proceedings
PublisherSpringer Verlag
Pages9-16
Number of pages8
ISBN (Print)9783319111780
DOIs
Publication statusPublished - 2014
Event24th International Conference on Artificial Neural Networks, ICANN 2014 - Hamburg, Germany
Duration: 2014 Sept 152014 Sept 19

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8681 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other24th International Conference on Artificial Neural Networks, ICANN 2014
Country/TerritoryGermany
CityHamburg
Period14/9/1514/9/19

Keywords

  • S-CTRNN
  • recurrent neural network
  • variance estimation

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint

Dive into the research topics of 'Learning and recognition of multiple fluctuating temporal patterns using S-CTRNN'. Together they form a unique fingerprint.

Cite this