Matches in SemOpenAlex for { <https://semopenalex.org/work/W2910580498> ?p ?o ?g. }
- W2910580498 abstract "With the rapid increase of large-scale, real-world datasets, it becomes critical to address the problem of long-tailed data distribution (i.e., a few classes account for most of the data, while most classes are under-represented). Existing solutions typically adopt class re-balancing strategies such as re-sampling and re-weighting based on the number of observations for each class. In this work, we argue that as the number of samples increases, the additional benefit of a newly added data point will diminish. We introduce a novel theoretical framework to measure data overlap by associating with each sample a small neighboring region rather than a single point. The effective number of samples is defined as the volume of samples and can be calculated by a simple formula $(1-beta^{n})/(1-beta)$, where $n$ is the number of samples and $beta in [0,1)$ is a hyperparameter. We design a re-weighting scheme that uses the effective number of samples for each class to re-balance the loss, thereby yielding a class-balanced loss. Comprehensive experiments are conducted on artificially induced long-tailed CIFAR datasets and large-scale datasets including ImageNet and iNaturalist. Our results show that when trained with the proposed class-balanced loss, the network is able to achieve significant performance gains on long-tailed datasets." @default.
- W2910580498 created "2019-01-25" @default.
- W2910580498 creator A5018609918 @default.
- W2910580498 creator A5041567418 @default.
- W2910580498 creator A5041864624 @default.
- W2910580498 creator A5047506888 @default.
- W2910580498 creator A5052768778 @default.
- W2910580498 date "2019-01-16" @default.
- W2910580498 modified "2023-09-27" @default.
- W2910580498 title "Class-Balanced Loss Based on Effective Number of Samples" @default.
- W2910580498 cites W167016754 @default.
- W2910580498 cites W1686810756 @default.
- W2910580498 cites W1797268635 @default.
- W2910580498 cites W1861492603 @default.
- W2910580498 cites W1941659294 @default.
- W2910580498 cites W1988790447 @default.
- W2910580498 cites W1989684337 @default.
- W2910580498 cites W2063471322 @default.
- W2910580498 cites W2065890194 @default.
- W2910580498 cites W2097117768 @default.
- W2910580498 cites W2104933073 @default.
- W2910580498 cites W2107742354 @default.
- W2910580498 cites W2108598243 @default.
- W2910580498 cites W2117539524 @default.
- W2910580498 cites W2118978333 @default.
- W2910580498 cites W2119191234 @default.
- W2910580498 cites W2136903812 @default.
- W2910580498 cites W2145607950 @default.
- W2910580498 cites W2148143831 @default.
- W2910580498 cites W2153579005 @default.
- W2910580498 cites W2163605009 @default.
- W2910580498 cites W2194775991 @default.
- W2910580498 cites W2296043783 @default.
- W2910580498 cites W2339172597 @default.
- W2910580498 cites W2402144811 @default.
- W2910580498 cites W2440599146 @default.
- W2910580498 cites W2469885745 @default.
- W2910580498 cites W2597603852 @default.
- W2910580498 cites W2622263826 @default.
- W2910580498 cites W2732026016 @default.
- W2910580498 cites W2753211788 @default.
- W2910580498 cites W2753300133 @default.
- W2910580498 cites W2767106145 @default.
- W2910580498 cites W2767623047 @default.
- W2910580498 cites W2775447965 @default.
- W2910580498 cites W2781292787 @default.
- W2910580498 cites W2794467005 @default.
- W2910580498 cites W2797977484 @default.
- W2910580498 cites W2798381792 @default.
- W2910580498 cites W2867270703 @default.
- W2910580498 cites W2884561390 @default.
- W2910580498 cites W2963371670 @default.
- W2910580498 cites W2963703197 @default.
- W2910580498 cites W2964050365 @default.
- W2910580498 cites W2964137095 @default.
- W2910580498 cites W3118608800 @default.
- W2910580498 doi "https://doi.org/10.48550/arxiv.1901.05555" @default.
- W2910580498 hasPublicationYear "2019" @default.
- W2910580498 type Work @default.
- W2910580498 sameAs 2910580498 @default.
- W2910580498 citedByCount "55" @default.
- W2910580498 countsByYear W29105804982019 @default.
- W2910580498 countsByYear W29105804982020 @default.
- W2910580498 countsByYear W29105804982021 @default.
- W2910580498 countsByYear W29105804982023 @default.
- W2910580498 crossrefType "posted-content" @default.
- W2910580498 hasAuthorship W2910580498A5018609918 @default.
- W2910580498 hasAuthorship W2910580498A5041567418 @default.
- W2910580498 hasAuthorship W2910580498A5041864624 @default.
- W2910580498 hasAuthorship W2910580498A5047506888 @default.
- W2910580498 hasAuthorship W2910580498A5052768778 @default.
- W2910580498 hasBestOaLocation W29105804981 @default.
- W2910580498 hasConcept C106131492 @default.
- W2910580498 hasConcept C11413529 @default.
- W2910580498 hasConcept C121332964 @default.
- W2910580498 hasConcept C126838900 @default.
- W2910580498 hasConcept C140779682 @default.
- W2910580498 hasConcept C154945302 @default.
- W2910580498 hasConcept C183115368 @default.
- W2910580498 hasConcept C185592680 @default.
- W2910580498 hasConcept C198531522 @default.
- W2910580498 hasConcept C2524010 @default.
- W2910580498 hasConcept C2777212361 @default.
- W2910580498 hasConcept C2778755073 @default.
- W2910580498 hasConcept C28719098 @default.
- W2910580498 hasConcept C31972630 @default.
- W2910580498 hasConcept C33923547 @default.
- W2910580498 hasConcept C41008148 @default.
- W2910580498 hasConcept C43617362 @default.
- W2910580498 hasConcept C62520636 @default.
- W2910580498 hasConcept C71924100 @default.
- W2910580498 hasConcept C8642999 @default.
- W2910580498 hasConceptScore W2910580498C106131492 @default.
- W2910580498 hasConceptScore W2910580498C11413529 @default.
- W2910580498 hasConceptScore W2910580498C121332964 @default.
- W2910580498 hasConceptScore W2910580498C126838900 @default.
- W2910580498 hasConceptScore W2910580498C140779682 @default.
- W2910580498 hasConceptScore W2910580498C154945302 @default.
- W2910580498 hasConceptScore W2910580498C183115368 @default.
- W2910580498 hasConceptScore W2910580498C185592680 @default.