FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W1706338097> ?p ?o ?g. }

• W1706338097 abstract "Structured prediction tasks in machine learning involve the simultaneous prediction of multiple labels. This is typically done by maximizing a score function on the space of labels, which decomposes as a sum of pairwise elements, each depending on two specific labels. Intuitively, the more pairwise terms are used, the better the expected accuracy. However, there is currently no theoretical account of this intuition. This paper takes a significant step in this direction. We formulate the problem as classifying the vertices of a known graph $G=(V,E)$, where the vertices and edges of the graph are labelled and correlate semi-randomly with the ground truth. We show that the prospects for achieving low expected Hamming error depend on the structure of the graph $G$ in interesting ways. For example, if $G$ is a very poor expander, like a path, then large expected Hamming error is inevitable. Our main positive result shows that, for a wide class of graphs including 2D grid graphs common in machine vision applications, there is a polynomial-time algorithm with small and information-theoretically near-optimal expected error. Our results provide a first step toward a theoretical justification for the empirical success of the efficient approximate inference algorithms that are used for structured prediction in models where exact inference is intractable." @default.
• W1706338097 created "2016-06-24" @default.
• W1706338097 creator A5013431623 @default.
• W1706338097 creator A5022886322 @default.
• W1706338097 creator A5047817959 @default.
• W1706338097 creator A5071230826 @default.
• W1706338097 date "2014-09-19" @default.
• W1706338097 modified "2023-09-27" @default.
• W1706338097 title "Tight Error Bounds for Structured Prediction." @default.
• W1706338097 cites W15368807 @default.
• W1706338097 cites W1605711022 @default.
• W1706338097 cites W1903393809 @default.
• W1706338097 cites W1971680255 @default.
• W1706338097 cites W1987178221 @default.
• W1706338097 cites W2008652694 @default.
• W1706338097 cites W2013520638 @default.
• W1706338097 cites W2016832592 @default.
• W1706338097 cites W2019441424 @default.
• W1706338097 cites W2024128809 @default.
• W1706338097 cites W2041504706 @default.
• W1706338097 cites W2057396406 @default.
• W1706338097 cites W2059017103 @default.
• W1706338097 cites W2060990051 @default.
• W1706338097 cites W2105644991 @default.
• W1706338097 cites W2113239207 @default.
• W1706338097 cites W2124037793 @default.
• W1706338097 cites W2128020980 @default.
• W1706338097 cites W2139686264 @default.
• W1706338097 cites W2146807782 @default.
• W1706338097 cites W2147880316 @default.
• W1706338097 cites W2147899407 @default.
• W1706338097 cites W2153396823 @default.
• W1706338097 cites W2158349948 @default.
• W1706338097 cites W2167026441 @default.
• W1706338097 cites W2519716980 @default.
• W1706338097 cites W2798588639 @default.
• W1706338097 cites W2914959486 @default.
• W1706338097 cites W2962923656 @default.
• W1706338097 hasPublicationYear "2014" @default.
• W1706338097 type Work @default.
• W1706338097 sameAs 1706338097 @default.
• W1706338097 citedByCount "8" @default.
• W1706338097 countsByYear W17063380972015 @default.
• W1706338097 countsByYear W17063380972017 @default.
• W1706338097 countsByYear W17063380972019 @default.
• W1706338097 crossrefType "posted-content" @default.
• W1706338097 hasAuthorship W1706338097A5013431623 @default.
• W1706338097 hasAuthorship W1706338097A5022886322 @default.
• W1706338097 hasAuthorship W1706338097A5047817959 @default.
• W1706338097 hasAuthorship W1706338097A5071230826 @default.
• W1706338097 hasConcept C111472728 @default.
• W1706338097 hasConcept C11413529 @default.
• W1706338097 hasConcept C132010649 @default.
• W1706338097 hasConcept C132525143 @default.
• W1706338097 hasConcept C138885662 @default.
• W1706338097 hasConcept C146849305 @default.
• W1706338097 hasConcept C154945302 @default.
• W1706338097 hasConcept C157125643 @default.
• W1706338097 hasConcept C184898388 @default.
• W1706338097 hasConcept C193319292 @default.
• W1706338097 hasConcept C2776214188 @default.
• W1706338097 hasConcept C33923547 @default.
• W1706338097 hasConcept C41008148 @default.
• W1706338097 hasConcept C57273362 @default.
• W1706338097 hasConcept C73150493 @default.
• W1706338097 hasConcept C80444323 @default.
• W1706338097 hasConceptScore W1706338097C111472728 @default.
• W1706338097 hasConceptScore W1706338097C11413529 @default.
• W1706338097 hasConceptScore W1706338097C132010649 @default.
• W1706338097 hasConceptScore W1706338097C132525143 @default.
• W1706338097 hasConceptScore W1706338097C138885662 @default.
• W1706338097 hasConceptScore W1706338097C146849305 @default.
• W1706338097 hasConceptScore W1706338097C154945302 @default.
• W1706338097 hasConceptScore W1706338097C157125643 @default.
• W1706338097 hasConceptScore W1706338097C184898388 @default.
• W1706338097 hasConceptScore W1706338097C193319292 @default.
• W1706338097 hasConceptScore W1706338097C2776214188 @default.
• W1706338097 hasConceptScore W1706338097C33923547 @default.
• W1706338097 hasConceptScore W1706338097C41008148 @default.
• W1706338097 hasConceptScore W1706338097C57273362 @default.
• W1706338097 hasConceptScore W1706338097C73150493 @default.
• W1706338097 hasConceptScore W1706338097C80444323 @default.
• W1706338097 hasLocation W17063380971 @default.
• W1706338097 hasOpenAccess W1706338097 @default.
• W1706338097 hasPrimaryLocation W17063380971 @default.
• W1706338097 hasRelatedWork W1619721866 @default.
• W1706338097 hasRelatedWork W1803708361 @default.
• W1706338097 hasRelatedWork W1954018497 @default.
• W1706338097 hasRelatedWork W1989042829 @default.
• W1706338097 hasRelatedWork W2004531067 @default.
• W1706338097 hasRelatedWork W2016776056 @default.
• W1706338097 hasRelatedWork W2024281442 @default.
• W1706338097 hasRelatedWork W2102907934 @default.
• W1706338097 hasRelatedWork W2114244473 @default.
• W1706338097 hasRelatedWork W2126641717 @default.
• W1706338097 hasRelatedWork W2127048411 @default.
• W1706338097 hasRelatedWork W2133361319 @default.
• W1706338097 hasRelatedWork W2143133882 @default.
• W1706338097 hasRelatedWork W2519716980 @default.
• W1706338097 hasRelatedWork W2947361127 @default.

• next