Combining Induction and Transduction for Abstract Reasoning
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916502208249856 |
|---|---|
| author | Li, Wen-Ding Hu, Keya Larsen, Carter Wu, Yuqing Alford, Simon Woo, Caleb Dunn, Spencer M. Tang, Hao Naim, Michelangelo Nguyen, Dat Zheng, Wei-Long Tavares, Zenna Pu, Yewen Ellis, Kevin |
| author_facet | Li, Wen-Ding Hu, Keya Larsen, Carter Wu, Yuqing Alford, Simon Woo, Caleb Dunn, Spencer M. Tang, Hao Naim, Michelangelo Nguyen, Dat Zheng, Wei-Long Tavares, Zenna Pu, Yewen Ellis, Kevin |
| contents | When learning an input-output mapping from very few examples, is it better to first infer a latent function that explains the examples, or is it better to directly predict new test outputs, e.g. using a neural network? We study this question on ARC by training neural models for induction (inferring latent functions) and transduction (directly predicting the test output for a given test input). We train on synthetically generated variations of Python programs that solve ARC training tasks. We find inductive and transductive models solve different kinds of test problems, despite having the same training problems and sharing the same neural architecture: Inductive program synthesis excels at precise computations, and at composing multiple concepts, while transduction succeeds on fuzzier perceptual concepts. Ensembling them approaches human-level performance on ARC. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2411_02272 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Combining Induction and Transduction for Abstract Reasoning Li, Wen-Ding Hu, Keya Larsen, Carter Wu, Yuqing Alford, Simon Woo, Caleb Dunn, Spencer M. Tang, Hao Naim, Michelangelo Nguyen, Dat Zheng, Wei-Long Tavares, Zenna Pu, Yewen Ellis, Kevin Machine Learning Artificial Intelligence Computation and Language When learning an input-output mapping from very few examples, is it better to first infer a latent function that explains the examples, or is it better to directly predict new test outputs, e.g. using a neural network? We study this question on ARC by training neural models for induction (inferring latent functions) and transduction (directly predicting the test output for a given test input). We train on synthetically generated variations of Python programs that solve ARC training tasks. We find inductive and transductive models solve different kinds of test problems, despite having the same training problems and sharing the same neural architecture: Inductive program synthesis excels at precise computations, and at composing multiple concepts, while transduction succeeds on fuzzier perceptual concepts. Ensembling them approaches human-level performance on ARC. |
| title | Combining Induction and Transduction for Abstract Reasoning |
| topic | Machine Learning Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2411.02272 |