Taming Binarized Neural Networks and Mixed-Integer Programs

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F24%3A00379059" target="_blank" >RIV/68407700:21230/24:00379059 - isvavai.cz</a>

Výsledek na webu

<a href="https://doi.org/10.1609/aaai.v38i10.28968" target="_blank" >https://doi.org/10.1609/aaai.v38i10.28968</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1609/aaai.v38i10.28968" target="_blank" >10.1609/aaai.v38i10.28968</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Taming Binarized Neural Networks and Mixed-Integer Programs

Popis výsledku v původním jazyce

There has been a great deal of recent interest in binarized neural networks, especially because of their explainability. At the same time, automatic differentiation algorithms such as back-propagation fail for binarized neural networks, which limits their applicability. We show that binarized neural networks admit a tame representation by reformulating the problem of training binarized neural networks as a subadditive dual of a mixed-integer program, which we show to have nice properties. This makes it possible to use the framework of Bolte et al. for implicit differentiation, which offers the possibility for practical implementation of backpropagation in the context of binarized neural networks. This approach could also be used for a broader class of mixed-integer programs, beyond the training of binarized neural networks, as encountered in symbolic approaches to AI and beyond.

Název v anglickém jazyce

Taming Binarized Neural Networks and Mixed-Integer Programs

Popis výsledku anglicky

There has been a great deal of recent interest in binarized neural networks, especially because of their explainability. At the same time, automatic differentiation algorithms such as back-propagation fail for binarized neural networks, which limits their applicability. We show that binarized neural networks admit a tame representation by reformulating the problem of training binarized neural networks as a subadditive dual of a mixed-integer program, which we show to have nice properties. This makes it possible to use the framework of Bolte et al. for implicit differentiation, which offers the possibility for practical implementation of backpropagation in the context of binarized neural networks. This approach could also be used for a broader class of mixed-integer programs, beyond the training of binarized neural networks, as encountered in symbolic approaches to AI and beyond.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

—

Návaznosti

R - Projekt Ramcoveho programu EK

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Proceedings of the 38th AAAI Conference on Artificial Intelligence

ISBN

—

ISSN

2159-5399

e-ISSN

2374-3468

Počet stran výsledku

9

Strana od-do

10935-10943

Název nakladatele

AAAI Press

Místo vydání

Menlo Park

Místo konání akce

Vancouver

Datum konání akce

20. 2. 2024

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

001241513600021