Learning World Models through Self Supervised Learning

Abstract

How do humans acquire the large amount of background knowledge about how the world works through observation in the first few weeks and months of life? Does this background knowledge constitute the basis of common sense? The working hypothesis is that the learning paradigm that could reproduce this ability in machines is self supervised learning (SSL).

Document Details

Document Type
DoD Grant Award
Publication Date
Jan 14, 2022
Source ID
FA95501910343

Entities

People

  • Yann Le Cun

Organizations

  • Air Force Office of Scientific Research
  • New York University
  • United States Air Force

Tags

Readers

  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Agricultural Chemistry/Soil Science
  • Theoretical Analysis.

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks