Hysteresis studies in a noisy autoassociative neural network

Area of Science:

• Computational neuroscience
• Complex systems

Background:

• Autoassociative neural networks (ANNs) are models of associative memory.
• Understanding ANN dynamics under external periodic driving is crucial for their application.

Purpose of the Study:

• To investigate magnetization dynamics in a noisy ANN under periodic external fields.
• To analyze the influence of drive amplitude, frequency, and noise on hysteresis loop area.

Main Methods:

• Numerical simulations of a noisy autoassociative neural network.
• Analysis of magnetization and hysteresis loop area as a function of system parameters.

Main Results:

• Hysteresis loop area shows a maximum at a specific nonzero noise intensity for weak periodic signals.
• This maximum indicates optimal synchronization between the periodic signal and the network's response.
• Hysteresis loop area also exhibits a maximum with respect to signal frequency.

Conclusions:

• Noise can enhance signal synchronization and network response in driven ANNs.
• The findings provide insights into the role of noise in information processing within neural networks.