SAI Simulator for Social AI Gossiping
Simulator developed within Humane-AI-Net Micro-Project "Social AI gossiping"
Main Characteristic
Python simulator.
Research areas
Collaborative AI
Technical Categories
Machine learning
Last updated
01.06.2023 - 15:47
Detailed Description
The SAI Simulator (SAISim) is a Python simulator for social-aware fully decentralized learning. It leverages state-of-the-art libraries such as PyTorch and NetworkX for deep learning and social graph creation, respectively. On top of that, SAISim implements the primitives for supporting fully decentralized learning. The local models of nodes are Multilayer Perceptrons with three layers (sizes 512, 256, 128) and ReLu activation functions. SGD is used for the optimization, with learning rate 0.001 and momentum 0.5.