Abstract:Nearly all models in neural networks start from the assumption that the input-output characteristic is a sigmoidal function. On parameter space, we present a systematic and feasible method for analyzing the whole spectrum of attractors—all-saturated, all-but-one-saturated, all-but-twosaturated, and so on—of a neurodynamical system with a saturated sigmoidal function as its input-output characteristic. We present an argument that claims, under a mild condition, that only all-saturated or all but-one-saturated attractors are observable for the neurodynamics. For any given all-saturated configuration [Formula: see text] (all-but-one-saturated configuration [Formula: see text]) the article shows how to construct an exact parameter region R([Formula: see text])([Formula: see text]([Formula: see text])) such that if and only if the parameters fall within R([Formula: see text])([Formula: see text]([Formula: see text])), then [Formula: see text]([Formula: see text]) is an attractor (a fixed point) of the dynamics. The parameter region for an all-saturated fixed-point attractor is independent of the specific choice of a saturated sigmoidal function, whereas for an all-but-one-saturated fixed point, it is sensitive to the input-output characteristic. Based on a similar idea, the role of weight normalization realized by a saturated sigmoidal function in competitive learning is discussed. A necessary and sufficient condition is provided to distinguish two kinds of competitive learning: stable competitive learning with the weight vectors representing extremes of input space and being fixed-point attractors, and unstable competitive learning. We apply our results to Linsker's model and (using extreme value theory in statistics) the Hopfield model and obtain some novel results on these two models.

Attractors and the Attraction Basins of Discrete-Time Cellular Neural Networks

Exponential attractor of κ-almost periodic sequence solution of discrete-time bidirectional neural networks

The Existence and Global Attractivity of Almost Periodic Sequence Solution of Discrete-Time Neural Networks

Nontrivial Global Attractors in 2-D Multistable Attractor Neural Networks

Continuous Attractors of Discrete-Time Recurrent Neural Networks

Analysis of Continuous Attractors for 2-D Linear Threshold Neural Networks

Representations Of Continuous Attractors Of Discrete-Time Cellular Neural Networks

A New Type of Chaotic Attractor with Cellular Neural Networks

Activity Invariant Sets and Exponentially Stable Attractors of Linear Threshold Discrete-Time Recurrent Neural Networks

Existence of Periodic Solutions and Closed Invariant Curves in a Class of Discrete-Time Cellular Neural Networks

On Attracting Basins of Multiple Equilibria of a Class of Cellular Neural Networks

Continuous Attractors and Population Decoding Multiple-Peaked Activity

Dynamics of Adaptive Continuous Attractor Neural Networks

Partially Unstable Attractors in Networks of Forced Integrate-and-fire Oscillators

Multistability and Attraction Basins of Discrete-Time Neural Networks with Nonmonotonic Piecewise Linear Activation Functions.

Dynamics and computation of continuous attractors.

Evaluation of Attractors and Basins of Asynchronous Random Boolean Networks

Fixed-Point Attractor Analysis for a Class of Neurodynamics

Strange Nonchaotic Attractor In High-Dimensional Neural System

Chaos in the discrete time cellular neural networks

Slow and Weak Attractor Computation Embedded in Fast and Strong E-I Balanced Neural Dynamics.