K2

K² · Artificial intelligence

To guess a missing machine part, look at where parts touch

The authors built Linkify, a framework for learning from interface-augmented assembly graphs to enable context-aware part retrieval in mechanical assemblies.

AJ
DG

Anushrut Jignasu, Daniele Grandi

2 authors · cs.CV

arXiv preprintArtificial intelligenceJul 2026 · ~75s read

The 30-second scan

Take apart any machine and you'll find the real story isn't in the parts themselves — it's in where they meet.

  1. They recomputed high-fidelity interface geometry for the Fusion 360 Gallery Assembly dataset, correcting missing and erroneous contacts and generating point-cloud representations of local contact regions.
  2. They constructed assembly graphs whose nodes encode part geometry and whose edges encode interface geometry via a pretrained point-cloud encoder.
  3. They trained a Graph Attention Network based on GATv2 to solve a masked part prediction task, predicting the class of a held-out component from a large vocabulary of geometrically clustered parts.
Read the originalPDF