PlaNeural: spiking neural networks that plan

Mitchell, Ian, Huyck, Christian R. and Evans, Carl (2016) PlaNeural: spiking neural networks that plan. In: 7th Annual International Conference on Biologically Inspired Cognitive Architectures, BICA 2016, 16 Jul 2016, New York City, NY, USA.

[img]
Preview
PDF - Published version (with publisher's formatting)
Available under License Creative Commons Attribution-NonCommercial-NoDerivatives.

Download (381kB) | Preview

Abstract

PlaNeural is a spike-based neural network that has the ability to plan. The network is a spreading activation network implemented with Cell Assemblies; this combination has built a dynamic network of nodes that is able to interact with an environment and respond appropriately. PlaNeural uses Cell Assemblies to make decisions and plan - there is no pre-determined code managing the decision process that leads to planning. PlaNeural is the planning component of a virtual robot in a virtual environment. This paper describes PlaNeural's behaviour in two virtual environments, programmed independently of it; actions are completed in a closed-loop. PlaNeural was programmed in PyNN, executed with Nest and on a neuromorphic platform, SpiNNaker. PlaNeural has been tested on two environments and results show a successful performance; in both cases PlaNeural takes appropriate actions to fulfil user selected goals based on environmental changes.

Item Type: Conference or Workshop Item (Paper)
Additional Information: Ian Mitchell, Christian Huyck, Carl Evans, PlaNeural: Spiking Neural Networks that Plan, Procedia Computer Science, Volume 88, 2016, Pages 198-204, ISSN 1877-0509, http://dx.doi.org/10.1016/j.procs.2016.07.425.
Research Areas: A. > School of Science and Technology > Computer Science > Artificial Intelligence group
Item ID: 20790
Notes on copyright: Selection and peer-review under responsibility of the Scientific Programme Committee of BICA 2016 © The Authors. Published by Elsevier B.V.
Useful Links:
Depositing User: Ian Mitchell
Date Deposited: 25 Oct 2016 09:38
Last Modified: 04 Apr 2019 06:01
URI: https://eprints.mdx.ac.uk/id/eprint/20790

Actions (login required)

Edit Item Edit Item

Full text downloads (NB count will be zero if no full text documents are attached to the record)

Downloads per month over the past year