Tropical Geometry of Deep Neural Networks · HackerPapers discussions

Tropical Geometry of Deep Neural Networks — HN submission

Sat, 09 Mar 2024 17:18:56 GMT

Tropical Geometry of Deep Neural Networks

1 point; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Tropical Geometry of Deep Neural Networks — mentioned by ganzuul

Wed, 21 Feb 2024 11:50:04 GMT

The feedback is like an exponential function fed to a ReLU. https://arxiv.org/abs/1805.07091 It was predictable that they hammered out what was impressive about it by trying to improve it with fast iteration towards a set of divergent goals.

Paper and discussion history. A mention is not an endorsement.

Tropical Geometry of Deep Neural Networks — mentioned by jesuslop

Thu, 24 Jan 2019 17:45:12 GMT

Really agree with the extra simplicity in the ReLU case. Liwen Zhang, Gregory Naitzat and Lek-Heng Lim showed last year that "the family of such neural networks is equivalent to the family of tropical rational maps", where rational functions are quotient of polynomials and "tropical" is in the context of Tropical Geometry, where instead of the typical "plus, times" ring one uses a "min, plus" ring that has somewhat unexpected applications. For instance, with its module theory one can calculate minimum costs paths in graphs just as one does for reachability starting with the adjacency matrix…

Paper and discussion history. A mention is not an endorsement.

Tropical Geometry of Deep Neural Networks — HN submission

Mon, 23 Jul 2018 08:56:19 GMT

Tropical Geometry of Deep Neural Networks

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.