Generalisable Agents for Neural Network Optimisation

arXiv:2311.18598v2 Announce Type: replace
Abstract: Optimising deep neural networks is a challenging task due to complex training dynamics, high computational requirements, and long training times. To address this difficulty, we propose the framework of Generalisable Agents for Neural Network Optimisation (GANNO) -- a multi-agent reinforcement learning (MARL) approach that learns to improve neural network optimisation by dynamically and responsively scheduling hyperparameters during training. GANNO utilises an agent per layer that observes localised network dynamics and accordingly takes actions to adjust …

abstract agent agents arxiv computational cs.ai cs.lg cs.ma dynamics framework multi-agent network networks neural network neural networks optimisation reinforcement reinforcement learning requirements training type