Multifunctionality in a reservoir computer
From MaRDI portal
Abstract: Multifunctionality is a well observed phenomenological feature of biological neural networks and considered to be of fundamental importance to the survival of certain species over time. These multifunctional neural networks are capable of performing more than one task without changing any network connections. In this paper we investigate how this neurological idiosyncrasy can be achieved in an artificial setting with a modern machine learning paradigm known as `Reservoir Computing'. A training technique is designed to enable a Reservoir Computer to perform tasks of a multifunctional nature. We explore the critical effects that changes in certain parameters can have on the Reservoir Computers' ability to express multifunctionality. We also expose the existence of several `untrained attractors'; attractors which dwell within the prediction state space of the Reservoir Computer that were not part of the training. We conduct a bifurcation analysis of these untrained attractors and discuss the implications of our results.
Recommendations
Cites work
- An experimental unification of reservoir computing methods
- Analysis of a new three-dimensional system with multiple chaotic attractors
- Chaotic attractors in delayed neural networks
- Control of multistability
- Controlling chaotic systems with multiple strange attractors
- Deterministic Nonperiodic Flow
- DEVELOPMENT OF THERAPEUTIC BRAIN STIMULATION TECHNIQUES WITH METHODS FROM NONLINEAR DYNAMICS AND STATISTICAL PHYSICS
- Real-Time Computing Without Stable States: A New Framework for Neural Computation Based on Perturbations
- Reservoir computing approaches to recurrent neural network training
- Using machine learning to replicate chaotic attractors and calculate Lyapunov exponents from data
Cited in
(10)- Symmetry kills the square in a multifunctional reservoir computer
- Functional identification of biological neural networks using reservoir adaptation for point processes
- Reservoir Computing with Complex Cellular Automata
- Forecasting macroscopic dynamics in adaptive Kuramoto network using reservoir computing
- Predicting collective states of a star network using reservoir computing
- SIGMa-DS: system identification from the geometric manifold of dynamical synchronization
- Confabulation dynamics in a reservoir computer: filling in the gaps with untrained attractors
- Pseudo-Lorenz attractors in echo state networks
- Statistics for differential topological properties between datasets with an application to reservoir computers
- How more data can hurt: instability and regularization in next-generation reservoir computing
This page was built for publication: Multifunctionality in a reservoir computer
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4983664)
