78389

Автор(ы): 

Автор(ов): 

1

Параметры публикации

Тип публикации: 

Глава в книге

Название: 

Direct and Inverse Problems of Time Encoding by Neuron-like Agents

Электронная публикация: 

Да

ISBN/ISSN: 

978-3-031-73691-9

DOI: 

10.1007/978-3-031-73691-9

Наименование источника: 

  • Advances in Neural Computation, Machine Learning, and Cognitive Research VIII

Город: 

  • Cham

Издательство: 

  • Springer Nature Switzerland AG

Год издания: 

2024

Страницы: 

353-361
Аннотация
In the paper, we present a model of a neuron-like agent with an internal pacemak-er network. Such agents communicate broadcastly through a common medium. They generate endogenous activity in definite time intervals, which are determined by their inner network. Every agent has a set of external and internal parameters. External parameters allow agents to interact with each other, and modulate (strengthen or weaken) their endogenous activity. The internal structure of an agent is represented by a network with three nodes: a half-center oscillator formed by two nodes with excitatory connections and a trigger, which is excited by both nodes of the oscillator. The trigger is capable of accumulating a potential using a memory mechanism. If the potential value exceeds the threshold value (one of the agent’s internal parameters), the trigger is activated. When activated, the trigger passes the activation to the agent. The agent's internal parameters, in addition to the threshold, are the parameters of its internal network: weights of connections, memory depth and forgetting rate. By changing these parameters, the agent can control the length of time intervals between two subsequent activations.

Библиографическая ссылка: 

Жилякова Л.Ю. Direct and Inverse Problems of Time Encoding by Neuron-like Agents / Advances in Neural Computation, Machine Learning, and Cognitive Research VIII. Cham: Springer Nature Switzerland AG, 2024. С. 353-361.