Empirical analysis of bifurcations in the full weights space of a two-neuron DTRNN
| Autor: | M. Gmez-Fuentes, J. Cervantes-Ojeda, R. Bernal-Jaquez |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
Discrete mathematics
Cognitive Neuroscience Structure (category theory) 02 engineering and technology Topology Space (mathematics) 01 natural sciences Computer Science Applications Bifurcation analysis Recurrent neural network Artificial Intelligence 0103 physical sciences Attractor 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Algebraic number Variety (universal algebra) 010301 acoustics Bifurcation Mathematics |
| Zdroj: | Neurocomputing. 237:362-374 |
| ISSN: | 0925-2312 |
| DOI: | 10.1016/j.neucom.2017.01.027 |
| Popis: | This is an empirical analysis of the dynamic behavior of Discrete-Time Recurrent Neural Networks (DTRNN) with two neurons based on the existing bifurcations on the full 4-dimensional synaptic weights space. We describe the existing bifurcation manifolds and the corresponding expected behavior in each region delimited by them in terms of the existing attractors. We found an unexpectedly rich variety of behaviors, however, finite and classifiable. We propose also an algebraic nomenclature that helps in the understanding of the underlying structure in the weights space. HighlightsBifurcation analysis of a two-neuron Discrete-Time Recurrent Neural Network.Coverage of the full 4-dimensional weights space separated into regions.Description of the expected behavior in each region in terms of existing attractors.Proposition of an algebraic language to describe the weights space structure. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect