Matches in SemOpenAlex for { <https://semopenalex.org/work/W3212909432> ?p ?o ?g. }
Showing items 1 to 68 of
68
with 100 items per page.
- W3212909432 abstract "We tackle sequential learning under label noise in applications where a human supervisor can be queried to relabel suspicious examples. Existing approaches are flawed, in that they only relabel incoming examples that look ``suspicious'' to the model. As a consequence, those mislabeled examples that elude (or don't undergo) this cleaning step end up tainting the training data and the model with no further chance of being cleaned. We propose Cincer, a novel approach that cleans both new and past data by identifying pairs of mutually incompatible examples. Whenever it detects a suspicious example, Cincer identifies a counter-example in the training set that -- according to the model -- is maximally incompatible with the suspicious example, and asks the annotator to relabel either or both examples, resolving this possible inconsistency. The counter-examples are chosen to be maximally incompatible, so to serve as explanations of the model' suspicion, and highly influential, so to convey as much information as possible if relabeled. Cincer achieves this by leveraging an efficient and robust approximation of influence functions based on the Fisher information matrix (FIM). Our extensive empirical evaluation shows that clarifying the reasons behind the model's suspicions by cleaning the counter-examples helps acquiring substantially better data and models, especially when paired with our FIM approximation." @default.
- W3212909432 created "2021-11-22" @default.
- W3212909432 creator A5001174680 @default.
- W3212909432 creator A5001227032 @default.
- W3212909432 creator A5030298064 @default.
- W3212909432 creator A5066187890 @default.
- W3212909432 date "2021-12-06" @default.
- W3212909432 modified "2023-09-24" @default.
- W3212909432 title "Interactive Label Cleaning with Example-based Explanations" @default.
- W3212909432 hasPublicationYear "2021" @default.
- W3212909432 type Work @default.
- W3212909432 sameAs 3212909432 @default.
- W3212909432 citedByCount "0" @default.
- W3212909432 crossrefType "proceedings-article" @default.
- W3212909432 hasAuthorship W3212909432A5001174680 @default.
- W3212909432 hasAuthorship W3212909432A5001227032 @default.
- W3212909432 hasAuthorship W3212909432A5030298064 @default.
- W3212909432 hasAuthorship W3212909432A5066187890 @default.
- W3212909432 hasConcept C107457646 @default.
- W3212909432 hasConcept C115961682 @default.
- W3212909432 hasConcept C119857082 @default.
- W3212909432 hasConcept C154945302 @default.
- W3212909432 hasConcept C177264268 @default.
- W3212909432 hasConcept C17744445 @default.
- W3212909432 hasConcept C199360897 @default.
- W3212909432 hasConcept C199539241 @default.
- W3212909432 hasConcept C2779110517 @default.
- W3212909432 hasConcept C41008148 @default.
- W3212909432 hasConcept C99498987 @default.
- W3212909432 hasConceptScore W3212909432C107457646 @default.
- W3212909432 hasConceptScore W3212909432C115961682 @default.
- W3212909432 hasConceptScore W3212909432C119857082 @default.
- W3212909432 hasConceptScore W3212909432C154945302 @default.
- W3212909432 hasConceptScore W3212909432C177264268 @default.
- W3212909432 hasConceptScore W3212909432C17744445 @default.
- W3212909432 hasConceptScore W3212909432C199360897 @default.
- W3212909432 hasConceptScore W3212909432C199539241 @default.
- W3212909432 hasConceptScore W3212909432C2779110517 @default.
- W3212909432 hasConceptScore W3212909432C41008148 @default.
- W3212909432 hasConceptScore W3212909432C99498987 @default.
- W3212909432 hasLocation W32129094321 @default.
- W3212909432 hasOpenAccess W3212909432 @default.
- W3212909432 hasPrimaryLocation W32129094321 @default.
- W3212909432 hasRelatedWork W2150384446 @default.
- W3212909432 hasRelatedWork W2778115935 @default.
- W3212909432 hasRelatedWork W2809037461 @default.
- W3212909432 hasRelatedWork W2809713126 @default.
- W3212909432 hasRelatedWork W2819137623 @default.
- W3212909432 hasRelatedWork W2897650178 @default.
- W3212909432 hasRelatedWork W2901810203 @default.
- W3212909432 hasRelatedWork W2910213810 @default.
- W3212909432 hasRelatedWork W2949479613 @default.
- W3212909432 hasRelatedWork W2950356213 @default.
- W3212909432 hasRelatedWork W2962939608 @default.
- W3212909432 hasRelatedWork W2971470662 @default.
- W3212909432 hasRelatedWork W3003178379 @default.
- W3212909432 hasRelatedWork W3082739451 @default.
- W3212909432 hasRelatedWork W3091808893 @default.
- W3212909432 hasRelatedWork W3163265547 @default.
- W3212909432 hasRelatedWork W3168133108 @default.
- W3212909432 hasRelatedWork W3185107419 @default.
- W3212909432 hasRelatedWork W3206275068 @default.
- W3212909432 hasRelatedWork W40872973 @default.
- W3212909432 hasVolume "34" @default.
- W3212909432 isParatext "false" @default.
- W3212909432 isRetracted "false" @default.
- W3212909432 magId "3212909432" @default.
- W3212909432 workType "article" @default.