Multiscale Generative Models: Improving Performance of a Generative Model Using Feedback from Other Dependent Generative Models

Authors: Changyu Chen, Avinandan Bose, Shih-Fen Cheng, Arunesh Sinha

Multiscale Generative Models (MGM) framework introduces an approach to train generative models with limited data in the context of multiagent systems. The agent with limited data receives feedbacks from other dependent agents, which essential results in a constrained generative model, i.e. its generative behavior is constrained by the feedbacks.

This repository contains an implementation and further details of MGM.

Reference: Changyu Chen, Avinandan Bose, Shih-Fen Cheng, Arunesh Sinha, "Multiscale generative models: Improving performance of a generative model using feedback from other dependent generative models." AAAI Conference on Artificial Intelligence (AAAI), 2022.

Paper Link: https://arxiv.org/abs/2201.09644

Contact: cychen.2020@phdcs.smu.edu.sg