Parameter calibration with Consensus-based Optimization for interaction dynamics driven by neural networks

TL;DR

This paper introduces a parameter calibration method using Consensus-based Optimization for neural network models of interaction dynamics, validated on traffic data and compared to traditional gradient descent methods.

Contribution

The paper presents a novel calibration approach with CBO for neural network interaction models, demonstrating its effectiveness on real traffic data.

Findings

CBO-based calibration produces forces comparable to physical models.

The method outperforms stochastic gradient descent in calibration accuracy.

Validated on real traffic data from ESIMAS experiment.

Abstract

We calibrate parameters of neural networks that model forces in interaction dynamics with the help of the Consensus-based global optimization method (CBO). We state the general framework of interaction particle systems driven by neural networks and test the proposed method with a real dataset from the ESIMAS traffic experiment. The resulting forces are compared to well-known physical interaction forces. Moreover, we compare the performance of the proposed calibration process to the one in [4] which uses a stochastic gradient descent algorithm.