Role of Noise in the Neuronal Ability to Encode and Classify Signals

Location

CSU Ballroom

Start Date

12-4-2022 10:00 AM

End Date

12-4-2022 11:30 AM

Student's Major

Physics and Astronomy

Student's College

Science, Engineering and Technology

Mentor's Name

Jorge Mendez

Mentor's Department

Physics and Astronomy

Mentor's College

Science, Engineering and Technology

Description

The brain is an intrinsically noisy environment. Neurons and networks are able to detect and classify different natural signals. The specific role played by noise in the codification of information is broadly unknown. The common picture of noise as a factor that can only cause a deterioration of the information in the signal is coming from linear systems. Most of the neurons in the brain and the mathematical models to represent them operate as excitable systems, which are nonlinear systems. Once nonlinearities are considered, noise can improve the codification of the signals. These improvements are usually called stochastic resonances. The first described stochastic resonance was obtained as an improvement in the response of a nonlinear system to a periodic input signal. This idea was later extended to more naturalistic relevant signals (aperiodic input signals), and referred as aperiodic stochastic resonance. In this study, we explore the significance of aperiodic stochastic resonance for the codification and classification of chaotic signals by a single neuron modeled using well established neuronal mathematical models. We analyze different metrics to quantify stochastic resonances, and the connection to a machine learning classification problem.

This document is currently not available here.

Share

COinS
 
Apr 12th, 10:00 AM Apr 12th, 11:30 AM

Role of Noise in the Neuronal Ability to Encode and Classify Signals

CSU Ballroom

The brain is an intrinsically noisy environment. Neurons and networks are able to detect and classify different natural signals. The specific role played by noise in the codification of information is broadly unknown. The common picture of noise as a factor that can only cause a deterioration of the information in the signal is coming from linear systems. Most of the neurons in the brain and the mathematical models to represent them operate as excitable systems, which are nonlinear systems. Once nonlinearities are considered, noise can improve the codification of the signals. These improvements are usually called stochastic resonances. The first described stochastic resonance was obtained as an improvement in the response of a nonlinear system to a periodic input signal. This idea was later extended to more naturalistic relevant signals (aperiodic input signals), and referred as aperiodic stochastic resonance. In this study, we explore the significance of aperiodic stochastic resonance for the codification and classification of chaotic signals by a single neuron modeled using well established neuronal mathematical models. We analyze different metrics to quantify stochastic resonances, and the connection to a machine learning classification problem.

Recommended Citation

Zakariya, Mohamed. "Role of Noise in the Neuronal Ability to Encode and Classify Signals." Undergraduate Research Symposium, Mankato, MN, April 12, 2022.
https://cornerstone.lib.mnsu.edu/urs/2022/poster-session-01/17