English

Learning to solve arithmetic problems with a virtual abacus

Machine Learning 2023-01-18 v1 Artificial Intelligence

Abstract

Acquiring mathematical skills is considered a key challenge for modern Artificial Intelligence systems. Inspired by the way humans discover numerical knowledge, here we introduce a deep reinforcement learning framework that allows to simulate how cognitive agents could gradually learn to solve arithmetic problems by interacting with a virtual abacus. The proposed model successfully learn to perform multi-digit additions and subtractions, achieving an error rate below 1% even when operands are much longer than those observed during training. We also compare the performance of learning agents receiving a different amount of explicit supervision, and we analyze the most common error patterns to better understand the limitations and biases resulting from our design choices.

Keywords

Cite

@article{arxiv.2301.06870,
  title  = {Learning to solve arithmetic problems with a virtual abacus},
  author = {Flavio Petruzzellis and Ling Xuan Chen and Alberto Testolin},
  journal= {arXiv preprint arXiv:2301.06870},
  year   = {2023}
}

Comments

Accepted at Northern Lights Deep Learning Conference 2023

R2 v1 2026-06-28T08:13:25.161Z