arXiv Machine Learning July 17, 2026 By Sridhar Mahadevan

Learning in Infinitesimal Non-Compositional Sketches

Read the original on arXiv Machine Learning →

arXiv:2607. 15107v1 Announce Type: new Abstract: This paper develops a categorical framework -- Learning in Infinitesimal Non-Compositional Sketches (LINCS) -- as the repair of non-compositionality: failures of diagrams to factor through quotient sketches lifted to the tangent category setting.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

arXiv Machine Learning

Jul 23

On the Computational Complexity of Structural Generalization

arXiv:2607. 19573v1 Announce Type: cross Abstract: Structural generalization has been measured repeatedly by several benchmarks, yet it has never been formally defined.

By Zichao Wei

llms benchmarks

arXiv Machine Learning

Jun 9

Learning Transfers: Kan Extensions for Neural Invariants

arXiv:2606. 07627v1 Announce Type: new Abstract: Transfer learning presumes that a representation learned on source tasks carries structure that remains usable on related target tasks.

By Luciano Melodia

fine-tuning

arXiv Machine Learning

Sep 22

Exactness at Inference: A Representational Criterion for Out-of-Distribution Generalization

arXiv:2609.24942v1 Announce Type: new Abstract: A model generalizes outside its training distribution only when it computes a representation structurally equivalent to the generating mechanism, not a...

By Filipe Marinho Rocha, In\^es Dutra, V\'itor Santos Costa, Lu\'is Paulo Reis

rag

arXiv Machine Learning

Aug 7

Positive-Data Learning of Fixed-Observation Linear MCFGs from Working Binary Presentations

arXiv:2605. 11644v2 Announce Type: replace-cross Abstract: We study positive-data learning of languages admitting reduced working binary linear nondeleting multiple context-free grammar presentations of bounded fan-out.

By Takayuki Kuriyama

arXiv Machine Learning

Jun 17

Monotonic Kolmogorov-Arnold Networks: A Theoretical and Empirical Study of Monotonicity as an Inductive Bias

arXiv:2606. 17886v1 Announce Type: new Abstract: Monotonicity has been a long-running architectural inductive bias for neural networks, motivated by tabular, scientific, and economic settings where outputs are known to respond monotonically to certain inputs.

By Mikhail Krasnov, Carolina Fortuna, Bla\v{z} Bertalani\v{c}

benchmarks safety

Hugging Face Trending Papers

Jun 16

Monotonic Kolmogorov-Arnold Networks: A Theoretical and Empirical Study of Monotonicity as an Inductive Bias

Monotonicity has been a long-running architectural inductive bias for neural networks, motivated by tabular, scientific, and economic settings where outputs are known to respond monotonically to certain inputs. Existing approaches are MLP- or flow-based and lack per-edge functional transparency; the only Kolmogorov--Arnold Network (KAN) variant with monotonicity, MonoKAN, enforces the constraint only on a restricted parameter subset and requires a projection-style training procedure.

benchmarks safety

One email a morning, machine-written

Related stories

On the Computational Complexity of Structural Generalization

Learning Transfers: Kan Extensions for Neural Invariants

Exactness at Inference: A Representational Criterion for Out-of-Distribution Generalization

Positive-Data Learning of Fixed-Observation Linear MCFGs from Working Binary Presentations

Monotonic Kolmogorov-Arnold Networks: A Theoretical and Empirical Study of Monotonicity as an Inductive Bias

Monotonic Kolmogorov-Arnold Networks: A Theoretical and Empirical Study of Monotonicity as an Inductive Bias