Papers by Kenny Smith
Meta-Learning to Compositionally Generalize (2021.acl-long)
Copied to clipboard
| Challenge: | Existing studies show that neural networks struggle with compositional generalization . prior work asserts that there are fundamental differences between cognitive and connectionist architectures that make compositional globalization unlikely. |
| Approach: | They propose a meta-learning augmented version of supervised learning that optimizes for out-of-distribution generalization. |
| Outcome: | The proposed model improves generalization performance on COGS and SCAN datasets. |
Recursive numeral systems are highly regular and easy to process (2026.eacl-long)
Copied to clipboard
Ponrawee Prasertsom, Andrea Silvi, Jennifer Culbertson, Devdatt Dubhashi, Moa Johansson, Kenny Smith
| Challenge: | Existing studies on linguistic efficiency have focused on the systematicity of forms, a key property of natural language. |
| Approach: | They propose to incorporate regularity across sets of forms in studies of efficiency in language . they use the Minimum Description Length approach to measure regularity and processing complexity . |
| Outcome: | The proposed method shows that recursive numeral systems are more efficient with respect to regularity and processing complexity. |