Skip to main content
Search

Instituto Federal de Educação, Ciência e Tecnologia do Ceará

2 related articles

Subscribe to updates
  1. Explaining abstention, not just predictions: MINABRO computes minimum-size abductive explanations for linear models with a reject option

    Linear classifiers with a reject option abstain when evidence is insufficient, but no method previously guaranteed both faithful and minimum-size explanations for those abstentions. MINABRO keeps a greedy ordering for classified instances, yielding minimum-size abductive explanations in O(n log n) time over bounded feature domains, and formulates rejected instances as a 0–1 integer linear program. Across ten public datasets it ran faster than the sampling-based Anchors and the linear-programming AbLinRO, even while solving the NP-hard rejection form.

  2. arXiv

    FlatNSD Mitigates Over-squashing by Adjusting the Sheaf Rather than Rewiring the Graph, Leading on Long-range Benchmarks

    The work introduces sheaf effective resistance, reframing over-squashing from a graph-topology bottleneck into a quantity adjustable via the sheaf, and proves that for flat vector bundles the Jacobian-sense over-squashing sensitivity is upper bounded by a quantity related to sheaf effective resistance; the resulting FlatNSD implicitly lowers total sheaf effective resistance without altering graph topology, performing well on TreeNeighborsMatch, ECHO-Synth, and Graph Transfer stress tests.

You’re all caught up