technique

Meta-Learning for Compositionality

techniqueactiveprovisionalmeta-learning-for-compositionality-ec8f595e·1 events·first seen 2d ago

Aliases: Meta-Learning for Compositionality

Co-occurring entities

Fodor and Pylyshyn's Systematicity Challenge Still Stands Brenden Lake Zenon Pylyshyn Marco Baroni Jerry Fodor

More like this (12)

meta-learning compositional generalization MAML (Model-Agnostic Meta-Learning)Operads for compositional reasoning in LLMs compositional language emergence Operadic consistency: a label-free signal for compositional reasoning failures in LLMs Compositional Reasoning Depth Predicts Clinical AI Failure Multimodal Learning Backdoor Unlearning Generalization: A Path Toward the Removal of Unknown Triggers in LLMs Causally Evaluating the Learnability of Formal Language Tasks Multi-Task Learning Attention Amnesia in Hybrid LLMs: When CoT Fine-Tuning Breaks Long-Range Recall, and How to Fix It

Recent events (1)

5arXiv · cs.CL·2d ago·source ↗

Fodor and Pylyshyn's systematicity challenge to neural networks remains unmet, paper argues

A new arXiv preprint argues that recent claims that neural networks have met Fodor and Pylyshyn's systematicity challenge are premature. The authors specifically target Lake and Baroni's meta-learning for compositionality (MLC) protocol, showing it struggles with out-of-distribution rules and behaves unsystematically on many within-distribution problems. The paper concludes that the classical cognitive science challenge — that neural networks cannot explain systematic biconditional dependencies in language and thought — remains unresolved.

Evaluation and Benchmarking AI Safety Research Fodor and Pylyshyn's Systematicity Challenge Still Stands Brenden Lake Zenon Pylyshyn +3 more