Learning differentiable logic programs for abstract visual reasoning
Machine Learning · 2024-12 · 113 (11-12) · 8533-8584
DOI: 10.1007/s10994-024-06610-2 ↗Kurz gefasst
Fachartikel in der Zeitschrift Machine Learning, online erschienen am 26.10.2024. Verfasst von Hikaru Shindo und 3 weiteren Autor:innen. Die Arbeit wurde bisher 14-mal zitiert (Crossref) und nennt 109 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: SPAICER, TAILOR, AICO.
Abstract (Originalsprache)
Visual reasoning is essential for building intelligent agents that understand the world and perform problem-solving beyond perception. Differentiable forward reasoning has been developed to integrate reasoning with gradient-based machine learning paradigms. However, due to the memory intensity, most existing approaches do not bring the best of the expressivity of first-order logic, excluding a crucial ability to solve abstract visual reasoning , where agents need to perform reasoning by using analogies on abstract concepts in different scenarios. To overcome this problem, we propose NEUro-symbolic Message-pAssiNg reasoNer (NEUMANN) , which is a graph-based differentiable forward reasoner, passing messages in a memory-efficient manner and handling structured programs with functors. Moreover, we propose a computationally-efficient structure learning algorithm to perform explanatory program induction on complex visual scenes. To evaluate, in addition to conventional visual reasoning tasks, we propose a new task, visual reasoning behind-the-scenes , where agents need to learn abstract programs and then answer queries by imagining scenes that are not observed. We empirically demonstrate that NEUMANN solves visual reasoning tasks efficiently, outperforming neural, symbolic, and neuro-symbolic baselines.
Förderung
- SPAICER
- TAILOR
- AICO
- Technische Universität Darmstadt
Zitieren
APA
Shindo, H., Pfanschilling, V., Dhami, D. S., & Kersting, K. (2024). Learning differentiable logic programs for abstract visual reasoning. Machine Learning, 113(11-12), 8533-8584. https://doi.org/10.1007/s10994-024-06610-2
Vancouver
Shindo H, Pfanschilling V, Dhami DS, Kersting K. Learning differentiable logic programs for abstract visual reasoning. Machine Learning. 2024;113(11-12):8533-8584. doi:10.1007/s10994-024-06610-2
BibTeX
@article{Shindo2024,
author = {Shindo, Hikaru and Pfanschilling, Viktor and Dhami, Devendra Singh and Kersting, Kristian},
title = {Learning differentiable logic programs for abstract visual reasoning},
journal = {Machine Learning},
year = {2024},
volume = {113},
number = {11-12},
pages = {8533--8584},
doi = {10.1007/s10994-024-06610-2}
}
EUVIMED