Automatic differentiation (AD) is a family of techniques to compute gradients of general programs directly . Here, we explore the use of AD in the context of time-driven agent-based simulations . In traffic signal timing optimization problems with high input dimension, the gradient-based method has substantially superior performance . We demonstrate that theapproach enables gradient–based training of neural network-controlledsimulation entities embedded in the model logic . We study the fidelity and overhead of the differentiable models, as wellas the convergence speed and solution quality achieved by gradient-free methods compared to Gradient-based methods . We also show that the approach can be used to train neural networks using neural networks with neural networks in themodel logic. It also demonstrates that it can also be used for training neural networks. It can be useful to train networks with a neural network

Author(s) : Philipp Andelfinger

Links : PDF - Abstract

Code :
Coursera

Keywords : neural - based - gradient - networks - differentiable -

Leave a Reply

Your email address will not be published. Required fields are marked *