Pith. sign in

REVIEW 4 cited by

Tropical Geometry of Deep Neural Networks

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 1805.07091 v1 pith:XBHPBOBQ submitted 2018-05-18 cs.LG math.AGstat.ML

Tropical Geometry of Deep Neural Networks

classification cs.LG math.AGstat.ML
keywords networkstropicalneuralgeometrydeeperfamilyfeedforwardnetwork
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

We establish, for the first time, connections between feedforward neural networks with ReLU activation and tropical geometry --- we show that the family of such neural networks is equivalent to the family of tropical rational maps. Among other things, we deduce that feedforward ReLU neural networks with one hidden layer can be characterized by zonotopes, which serve as building blocks for deeper networks; we relate decision boundaries of such neural networks to tropical hypersurfaces, a major object of study in tropical geometry; and we prove that linear regions of such neural networks correspond to vertices of polytopes associated with tropical rational functions. An insight from our tropical formulation is that a deeper network is exponentially more expressive than a shallow network.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Tubular Neighbourhoods of Pfaffian Sets and Applications to Neural Networks

    math.AG 2026-07 conditional novelty 7.0

    Tube-volume bounds for smooth Pfaffian hypersurfaces yield condition-number tails for Pfaffian neural classifiers, with polynomial-in-width control for single-layer rational-weight sigmoids.

  2. TwistedMerge: Certified Higher-Order Diagnostics and Abstention for Model Merging

    cs.LG 2026-07 conditional novelty 6.0

    Cycle inconsistency in model merging is not automatically a cohomological obstruction: TwistedMerge certifies a class only after frozen-complex, centrality, closure, and statistical gates, and finds no natural central class.

  3. Lattice theory and algebraic models for deep convolutional learning based on mathematical morphology

    cs.AI 2026-05 unverdicted novelty 6.0

    Standard CNN pipeline is a cross-lattice non-idempotent operator; three idempotent morphological layer designs are characterized using lattice adjunctions.

  4. Detropicalization as a proof technique

    math.CO 2026-05 unverdicted novelty 5.0

    Evaluating rational functions without subtraction over the tropical semiring yields combinatorial interpretations that can serve as proofs.