Matches in SemOpenAlex for { <https://semopenalex.org/work/W3168607751> ?p ?o ?g. }
- W3168607751 abstract "Partial-label learning (PLL) utilizes instances with PLs, where a PL includes several candidate labels but only one is the true label (TL). In PLL, identification-based strategy (IBS) purifies each PL on the fly to select the (most likely) TL for training; average-based strategy (ABS) treats all candidate labels equally for training and let trained models be able to predict TL. Although PLL research has focused on IBS for better performance, ABS is also worthy of study since modern IBS behaves like ABS in the beginning of training to prepare for PL purification and TL selection. In this paper, we analyze why ABS was unsatisfactory and propose how to improve it. Theoretically, we formalize five problem settings of PLL and prove that average PL losses (APLLs) with bounded multi-class losses are always robust, while APLLs with unbounded losses may be non-robust, which is the first robustness analysis for PLL. Experimentally, we have two promising findings: ABS using bounded losses can match/exceed state-of-the-art performance of IBS using unbounded losses; after using robust APLLs to warm start, IBS can further improve upon itself. Our work draws attention to ABS research, which can in turn boost IBS and push forward the whole PLL." @default.
- W3168607751 created "2021-06-22" @default.
- W3168607751 creator A5007098537 @default.
- W3168607751 creator A5016620131 @default.
- W3168607751 creator A5017743551 @default.
- W3168607751 creator A5040537076 @default.
- W3168607751 creator A5072744508 @default.
- W3168607751 creator A5074742406 @default.
- W3168607751 creator A5075377647 @default.
- W3168607751 date "2021-06-10" @default.
- W3168607751 modified "2023-09-27" @default.
- W3168607751 title "On the Robustness of Average Losses for Partial-Label Learning" @default.
- W3168607751 cites W1866935739 @default.
- W3168607751 cites W1969623397 @default.
- W3168607751 cites W1975128126 @default.
- W3168607751 cites W1982032418 @default.
- W3168607751 cites W1994616650 @default.
- W3168607751 cites W2096072446 @default.
- W3168607751 cites W2102348129 @default.
- W3168607751 cites W2106008047 @default.
- W3168607751 cites W2107189314 @default.
- W3168607751 cites W2111478152 @default.
- W3168607751 cites W2112796928 @default.
- W3168607751 cites W2113290770 @default.
- W3168607751 cites W2120409953 @default.
- W3168607751 cites W2132509897 @default.
- W3168607751 cites W2144372981 @default.
- W3168607751 cites W2146111596 @default.
- W3168607751 cites W2158681777 @default.
- W3168607751 cites W2285223868 @default.
- W3168607751 cites W2393384312 @default.
- W3168607751 cites W2604553154 @default.
- W3168607751 cites W2733555913 @default.
- W3168607751 cites W2746791238 @default.
- W3168607751 cites W2750384547 @default.
- W3168607751 cites W2902986194 @default.
- W3168607751 cites W2903641274 @default.
- W3168607751 cites W2905443329 @default.
- W3168607751 cites W2963351448 @default.
- W3168607751 cites W2963446712 @default.
- W3168607751 cites W2963697299 @default.
- W3168607751 cites W2963735582 @default.
- W3168607751 cites W2963759070 @default.
- W3168607751 cites W2964292098 @default.
- W3168607751 cites W2964666459 @default.
- W3168607751 cites W2970971581 @default.
- W3168607751 cites W2971193222 @default.
- W3168607751 cites W2981873476 @default.
- W3168607751 cites W2995315671 @default.
- W3168607751 cites W2998135793 @default.
- W3168607751 cites W3035019293 @default.
- W3168607751 cites W3035436849 @default.
- W3168607751 cites W3101114756 @default.
- W3168607751 cites W3118608800 @default.
- W3168607751 cites W3129692029 @default.
- W3168607751 doi "https://doi.org/10.48550/arxiv.2106.06152" @default.
- W3168607751 hasPublicationYear "2021" @default.
- W3168607751 type Work @default.
- W3168607751 sameAs 3168607751 @default.
- W3168607751 citedByCount "0" @default.
- W3168607751 crossrefType "posted-content" @default.
- W3168607751 hasAuthorship W3168607751A5007098537 @default.
- W3168607751 hasAuthorship W3168607751A5016620131 @default.
- W3168607751 hasAuthorship W3168607751A5017743551 @default.
- W3168607751 hasAuthorship W3168607751A5040537076 @default.
- W3168607751 hasAuthorship W3168607751A5072744508 @default.
- W3168607751 hasAuthorship W3168607751A5074742406 @default.
- W3168607751 hasAuthorship W3168607751A5075377647 @default.
- W3168607751 hasBestOaLocation W31686077511 @default.
- W3168607751 hasConcept C104317684 @default.
- W3168607751 hasConcept C119857082 @default.
- W3168607751 hasConcept C12707504 @default.
- W3168607751 hasConcept C134306372 @default.
- W3168607751 hasConcept C134652429 @default.
- W3168607751 hasConcept C154945302 @default.
- W3168607751 hasConcept C185592680 @default.
- W3168607751 hasConcept C2775924081 @default.
- W3168607751 hasConcept C33923547 @default.
- W3168607751 hasConcept C34388435 @default.
- W3168607751 hasConcept C41008148 @default.
- W3168607751 hasConcept C47446073 @default.
- W3168607751 hasConcept C55493867 @default.
- W3168607751 hasConcept C63479239 @default.
- W3168607751 hasConcept C76155785 @default.
- W3168607751 hasConceptScore W3168607751C104317684 @default.
- W3168607751 hasConceptScore W3168607751C119857082 @default.
- W3168607751 hasConceptScore W3168607751C12707504 @default.
- W3168607751 hasConceptScore W3168607751C134306372 @default.
- W3168607751 hasConceptScore W3168607751C134652429 @default.
- W3168607751 hasConceptScore W3168607751C154945302 @default.
- W3168607751 hasConceptScore W3168607751C185592680 @default.
- W3168607751 hasConceptScore W3168607751C2775924081 @default.
- W3168607751 hasConceptScore W3168607751C33923547 @default.
- W3168607751 hasConceptScore W3168607751C34388435 @default.
- W3168607751 hasConceptScore W3168607751C41008148 @default.
- W3168607751 hasConceptScore W3168607751C47446073 @default.
- W3168607751 hasConceptScore W3168607751C55493867 @default.
- W3168607751 hasConceptScore W3168607751C63479239 @default.
- W3168607751 hasConceptScore W3168607751C76155785 @default.
- W3168607751 hasLocation W31686077511 @default.