Disentangling neural disjunctive normal form models
File(s)
Published version
Author(s)
Type
Conference Paper
Abstract
Neural Disjunctive Normal Form (DNF) based models are powerful and interpretable approaches to neuro-symbolic learning and have shown promising results in classification and reinforcement learning settings without prior knowledge of the tasks. However, their performance is degraded by the thresholding of the post-training symbolic translation process. We show here that part of the performance degradation during translation is due to its failure to disentangle the learned knowledge represented in the form of the networks’ weights. We address this issue by proposing a new disentanglement method; by splitting nodes that encode nested rules into smaller independent nodes, we are able to better preserve the models’ performance. Through experiments on binary, multiclass, and multilabel classification tasks (including those requiring predicate invention), we demonstrate that our disentanglement method provides compact and interpretable logical representations for the neural DNF-based models, with performance closer to that of their pre-translation counterparts. Our code is available at https: //github.com/kittykg/disentangling-ndnf-classification.
Date Issued
2025-09-08
Date Acceptance
2025-09-01
Citation
Proceedings of Machine Learning Research, 2025, 284, pp.463-493
ISSN
2640-3498
Publisher
MLResearchPress
Start Page
463
End Page
493
Journal / Book Title
Proceedings of Machine Learning Research
Volume
284
Copyright Statement
© 2025 K.G. Baugh, V. Perreault, M. Baugh, L. Dickens, K. Inoue & A. Russo.
Source
Conference on Neurosymbolic Learning and Reasoning
Publication Status
Published
Start Date
2025-09-08
Finish Date
2025-09-10
Coverage Spatial
Santa Cruz, CA, USA