1 / 8

Creating Artificial Life Using RNNs and Hebbian Learning

This project explores artificial life generation through recurrent neural networks (RNNs) and Hebbian learning. By deriving RNNs from genomes, organisms are created with a 360-degree visual system. The encoding methods, including graph grammar and cellular encoding, facilitate the generation of adjacency matrices that represent neural network structures. The organisms navigate a wrap-around grid world where they can see and consume food, utilizing their RNN to determine actions. While foundational concepts are implemented, several aspects and extensions of the model remain to be developed.

agalia
Télécharger la présentation

Creating Artificial Life Using RNNs and Hebbian Learning

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Artificial Life Using Recurrent Neural Networks and Hebbian Learning Michael Tauraso CS 152

  2. Organisms • Behavior decided by a Recurrent Neural Network (RNN) • RNN derived from a genome • Organisms have a 360 degree visual system

  3. Gene Encoding • Direct Encoding (boring) • Graph Grammar (interesting) • Cellular Encoding (Interesting) • Functional Block Problem

  4. Graph Grammar • Goal is to generate an adjacency matrix • Each non-terminal generates a 2x2 matrix • Each 2x2 matrix contains either numbers or other non-terminals. • The start symbol is used to build up an adjacency matrix

  5. Cellular Encoding • Fundamental unit of genome is a transformation. • Transformations act on a particular neuron • A sequence of transformations defines a neural network. • This preserves functional blocks better

  6. Grid World • World is a wrap-around grid • Each grid can have 1 organism • Each grid can have food on it for the organisms to eat • Organisms have a 360 degree visual system • Organisms use an RNN to decide on their actions

  7. Learning Method • Recurrent networks • Hebbian Learning • Clamping Outputs

  8. Results? • Grid World difficult to implement • Two Interesting encodings remain unimplemented • Interesting extensions to the world model remain unimplemented

More Related