Knowledge transfer between simple discrete worlds

Key points

• Adapting to a new world and reasoning correctly about which outcomes are possible under rules one has not met before is a basic capacity underlying general intelligence.1
• We study it in small discrete worlds: sets of grid configurations consistent with a shared set of rules, concretely Sudoku-like puzzles built from a common base and a handful of extra rules.1
• The model receives the active rules either as a symbolic description or as two solved examples, and a model that gets no rule information is used as a baseline.1
• The framework allows us to test many such generalization regimes in more systematic way than benchmarks such as ARC-AGI and provides new ground for evaluating depth-recurrent reasoning models, which are typically tested on saturated Sudoku benchmarks.1

Sources (1)

• [1]Knowledge transfer between simple discrete worlds
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 10, 05:28 PM

  We train a single depth-recurrent model on many combinations of rules and test it at four levels of novelty: a new puzzle, a new combination of known rules, a pairing of rules never seen together, and a rule instance not seen at all during training.

  Adapting to a new world and reasoning correctly about which outcomes are possible under rules one has not met before is a basic capacity underlying general intelligence.

Extractive summary: sentences quoted from the sources.