DeepAI AI Chat
Log In Sign Up

Learning with AMIGo: Adversarially Motivated Intrinsic Goals

06/22/2020
by   Andres Campero, et al.
MIT
Facebook
NYU college
22

A key challenge for reinforcement learning (RL) consists of learning in environments with sparse extrinsic rewards. In contrast to current RL methods, humans are able to learn new skills with little or no reward by using various forms of intrinsic motivation. We propose AMIGo, a novel agent incorporating a goal-generating teacher that proposes Adversarially Motivated Intrinsic Goals to train a goal-conditioned "student" policy in the absence of (or alongside) environment reward. Specifically, through a simple but effective "constructively adversarial" objective, the teacher learns to propose increasingly challenging—yet achievable—goals that allow the student to learn general skills for acting in a new environment, independent of the task to be solved. We show that our method generates a natural curriculum of self-proposed goals which ultimately allows the agent to solve challenging procedurally-generated tasks where other forms of intrinsic motivation and state-of-the-art RL methods fail.

READ FULL TEXT

page 2

page 6

page 8

page 14

page 17

12/17/2020

Intrinsically Motivated Goal-Conditioned Reinforcement Learning: a Short Survey

Building autonomous machines that can explore open-ended environments, d...
03/04/2022

AutoDIME: Automatic Design of Interesting Multi-Agent Environments

Designing a distribution of environments in which RL agents can learn in...
12/07/2021

Information is Power: Intrinsic Control via Information Capture

Humans and animals explore their environment and acquire useful skills e...
06/23/2019

Neural networks with motivation

Motivational salience is a mechanism that determines an organism's curre...
06/23/2020

Show me the Way: Intrinsic Motivation from Demonstrations

The study of exploration in Reinforcement Learning (RL) has a long histo...
04/20/2016

Hierarchical Deep Reinforcement Learning: Integrating Temporal Abstraction and Intrinsic Motivation

Learning goal-directed behavior in environments with sparse feedback is ...

Code Repositories

impact-driven-exploration

impact-driven-exploration


view repo

adversarially-motivated-intrinsic-goals

This repository contains code for the method and experiments of the paper "Learning with AMIGo: Adversarially Motivated Intrinsic Goals".


view repo

adversarially-motivated-intrinsic-goals-variation

None


view repo