ARC

Possible Solutions

• Probabilistic Programming Program Synthesis
  • Gen: https://probcomp.github.io/Gen/
  • Pyro
• Program Synthesis by Graph Deep Learning
  • AST with parameters
• Train program synthesis on previously-priors
  • Model knows how to draw, fill, resize
• RL Agent + Program Synthesizer + Simulator
  • Agent reads actual state and synthesize
  • Generate many prior knowledge by simulation
  • Tenenbaum's idea of having a simulator that we compare with reality we perceive

Papers

A thought is a program

• Refs

  • https://www.vicarious.com/2019/01/18/a-thought-is-a-program/
  • https://github.com/vicariousinc/science_cognitive_programs
  • https://slideslive.com/38909792/building-machines-that-work-like-the-brain

• How do we acquire, represent and infer concepts?

Roots

• Lisp / Scheme

Refs

• https://fchollet-kaggle.s3.amazonaws.com/data/Next+generation+AI+benchmark.pdf
• Inductive Programming
  • https://www.microsoft.com/en-us/research/wp-content/uploads/2016/12/ip-cacm15.pdf
• One of prof. Tenenbaum students: http://alexlew.net/
• https://techburst.io/deep-learning-est-mort-vive-differentiable-programming-5060d3c55074