Matches in SemOpenAlex for { <https://semopenalex.org/work/W3129059619> ?p ?o ?g. }
- W3129059619 abstract "A fundamental assumption of most machine learning algorithms is that the training and test data are drawn from the same underlying distribution. However, this assumption is violated in almost all practical applications: machine learning systems are regularly tested under distribution shift, due to changing temporal correlations, atypical end users, or other factors. In this work, we consider the setting where the training data are structured into groups and there may be multiple test time shifts, corresponding to new groups or group distributions. Most prior methods aim to learn a single robust model or invariant feature space to tackle this group shift. In contrast, we aim to learn models that adapt at test time to shift using unlabeled test points. Our primary contribution is to introduce the framework of adaptive risk minimization (ARM), in which models are optimized for post adaptation performance on training batches sampled from different groups, which simulate group shifts that may occur at test time. We use meta-learning to solve the ARM problem, and compared to prior methods for robustness, invariance, and adaptation, ARM methods provide consistent gains of 1-4% test accuracy on image classification problems exhibiting group shift." @default.
- W3129059619 created "2021-02-15" @default.
- W3129059619 creator A5000779244 @default.
- W3129059619 creator A5005431772 @default.
- W3129059619 creator A5026322200 @default.
- W3129059619 creator A5056012604 @default.
- W3129059619 creator A5062586036 @default.
- W3129059619 creator A5068429450 @default.
- W3129059619 date "2020-07-06" @default.
- W3129059619 modified "2023-09-25" @default.
- W3129059619 title "Adaptive Risk Minimization: A Meta-Learning Approach for Tackling Group Distribution Shift" @default.
- W3129059619 cites W1598377843 @default.
- W3129059619 cites W1909320841 @default.
- W3129059619 cites W1959608418 @default.
- W3129059619 cites W2034368206 @default.
- W3129059619 cites W2120354757 @default.
- W3129059619 cites W2148603752 @default.
- W3129059619 cites W2167366427 @default.
- W3129059619 cites W2194775991 @default.
- W3129059619 cites W2299668505 @default.
- W3129059619 cites W2472819217 @default.
- W3129059619 cites W2547875792 @default.
- W3129059619 cites W2593768305 @default.
- W3129059619 cites W2601450892 @default.
- W3129059619 cites W2753160622 @default.
- W3129059619 cites W2763549966 @default.
- W3129059619 cites W2768783441 @default.
- W3129059619 cites W2785542264 @default.
- W3129059619 cites W2785902771 @default.
- W3129059619 cites W2798658180 @default.
- W3129059619 cites W2894728917 @default.
- W3129059619 cites W2895106137 @default.
- W3129059619 cites W2895671740 @default.
- W3129059619 cites W2917041815 @default.
- W3129059619 cites W2949117887 @default.
- W3129059619 cites W2952165242 @default.
- W3129059619 cites W2952610664 @default.
- W3129059619 cites W2962908092 @default.
- W3129059619 cites W2963043696 @default.
- W3129059619 cites W2963060032 @default.
- W3129059619 cites W2963826681 @default.
- W3129059619 cites W2970119729 @default.
- W3129059619 cites W2970971581 @default.
- W3129059619 cites W2991497298 @default.
- W3129059619 cites W2995735216 @default.
- W3129059619 cites W3020429927 @default.
- W3129059619 cites W3039431315 @default.
- W3129059619 cites W3039883906 @default.
- W3129059619 cites W3104140305 @default.
- W3129059619 cites W3117088788 @default.
- W3129059619 cites W3118608800 @default.
- W3129059619 cites W3119543788 @default.
- W3129059619 cites W99485931 @default.
- W3129059619 hasPublicationYear "2020" @default.
- W3129059619 type Work @default.
- W3129059619 sameAs 3129059619 @default.
- W3129059619 citedByCount "11" @default.
- W3129059619 countsByYear W31290596192020 @default.
- W3129059619 countsByYear W31290596192021 @default.
- W3129059619 countsByYear W31290596192022 @default.
- W3129059619 crossrefType "posted-content" @default.
- W3129059619 hasAuthorship W3129059619A5000779244 @default.
- W3129059619 hasAuthorship W3129059619A5005431772 @default.
- W3129059619 hasAuthorship W3129059619A5026322200 @default.
- W3129059619 hasAuthorship W3129059619A5056012604 @default.
- W3129059619 hasAuthorship W3129059619A5062586036 @default.
- W3129059619 hasAuthorship W3129059619A5068429450 @default.
- W3129059619 hasConcept C104317684 @default.
- W3129059619 hasConcept C107321475 @default.
- W3129059619 hasConcept C119857082 @default.
- W3129059619 hasConcept C12267149 @default.
- W3129059619 hasConcept C147764199 @default.
- W3129059619 hasConcept C154507838 @default.
- W3129059619 hasConcept C154945302 @default.
- W3129059619 hasConcept C16910744 @default.
- W3129059619 hasConcept C185592680 @default.
- W3129059619 hasConcept C199360897 @default.
- W3129059619 hasConcept C2776502983 @default.
- W3129059619 hasConcept C41008148 @default.
- W3129059619 hasConcept C51632099 @default.
- W3129059619 hasConcept C55493867 @default.
- W3129059619 hasConcept C63479239 @default.
- W3129059619 hasConceptScore W3129059619C104317684 @default.
- W3129059619 hasConceptScore W3129059619C107321475 @default.
- W3129059619 hasConceptScore W3129059619C119857082 @default.
- W3129059619 hasConceptScore W3129059619C12267149 @default.
- W3129059619 hasConceptScore W3129059619C147764199 @default.
- W3129059619 hasConceptScore W3129059619C154507838 @default.
- W3129059619 hasConceptScore W3129059619C154945302 @default.
- W3129059619 hasConceptScore W3129059619C16910744 @default.
- W3129059619 hasConceptScore W3129059619C185592680 @default.
- W3129059619 hasConceptScore W3129059619C199360897 @default.
- W3129059619 hasConceptScore W3129059619C2776502983 @default.
- W3129059619 hasConceptScore W3129059619C41008148 @default.
- W3129059619 hasConceptScore W3129059619C51632099 @default.
- W3129059619 hasConceptScore W3129059619C55493867 @default.
- W3129059619 hasConceptScore W3129059619C63479239 @default.
- W3129059619 hasLocation W31290596191 @default.
- W3129059619 hasOpenAccess W3129059619 @default.
- W3129059619 hasPrimaryLocation W31290596191 @default.