Assur modules improve limited-data learning of planar mechanisms
Original title: Efficient Learning of Mechanism Designs and Compositionality
Assur-derived features give a geometric-mean 1.8× sample-efficiency gain over flattened adjacency and position inputs.
The paper asks whether reusable constraint subassemblies help predict planar mechanism behavior from limited data. Assur-KB encodes irreducible modules, boundary interfaces, and solve dependencies while keeping source geometry. Theory covers transfer through module interfaces and limits from missing interface data and accumulated sensitivity. Matched multilayer-perceptron experiments support the full representation in the tested setting.
- Training budgets range from 50 to 10,000 mechanisms.
- Tests report invariance under the joint relabelings that were tried.
- Dyadic composition yields higher valid-generation rates in separate experiments.
- A coarser structural index covers 10.02 million mechanisms.
From the original article
1 min read at arxiv.orgLearning mechanism behaviour requires a representation of how connected parts jointly constrain motion. We study whether exposing reusable constraint subassemblies improves prediction from limited data. Assur-KB represents planar mechanisms through irreducible modules, boundary interfaces, and solve dependencies while retaining source geometry.
By João Pedro Gandarela · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.