Zero-Temperature Limit of a Convergent Algorithm to Minimize the Bethe Free Energy

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F11%3A00187154" target="_blank" >RIV/68407700:21230/11:00187154 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Zero-Temperature Limit of a Convergent Algorithm to Minimize the Bethe Free Energy

Popis výsledku v původním jazyce

After the discovery that fixed points of loopy belief propagation coincide with stationary points of the Bethe free energy, several reseachers proposed provably convergent algorithms to directly minimize the Bethe free energy. These algorithms were formulated only for non-zero temperature (thus finding fixed points of the sum-product algorithm) and their possible extension to zero temperature is not obvious. We present the zero-temperature limit of the double-loop algorithm by Heskes, which converges amax-product fixed point. The inner loop of this algorithm turns out to be known as max-sum diffusion. Under certain conditions, the algorithm combines the complementary advantages of the max-product algorithm and max-sum diffusion: it yields good approximation of both ground states and max-marginals.

Název v anglickém jazyce

Zero-Temperature Limit of a Convergent Algorithm to Minimize the Bethe Free Energy

Popis výsledku anglicky

After the discovery that fixed points of loopy belief propagation coincide with stationary points of the Bethe free energy, several reseachers proposed provably convergent algorithms to directly minimize the Bethe free energy. These algorithms were formulated only for non-zero temperature (thus finding fixed points of the sum-product algorithm) and their possible extension to zero temperature is not obvious. We present the zero-temperature limit of the double-loop algorithm by Heskes, which converges amax-product fixed point. The inner loop of this algorithm turns out to be known as max-sum diffusion. Under certain conditions, the algorithm combines the complementary advantages of the max-product algorithm and max-sum diffusion: it yields good approximation of both ground states and max-marginals.

Druh

O - Ostatní výsledky

CEP obor

JD - Využití počítačů, robotika a její aplikace

OECD FORD obor

—

Projekt

<a href="/cs/project/GAP103%2F10%2F0783" target="_blank" >GAP103/10/0783: Struktura a její využití při rozpoznávání</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2011

Kód důvěrnosti údajů

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