FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3204195324> ?p ?o ?g. }

• W3204195324 abstract "Understanding the trustworthiness of a prediction yielded by a classifier is critical for the safe and effective use of AI models. Prior efforts have been proven to be reliable on small-scale datasets. In this work, we study the problem of predicting trustworthiness on real-world large-scale datasets, where the task is more challenging due to high-dimensional features, diverse visual concepts, and large-scale samples. In such a setting, we observe that the trustworthiness predictors trained with prior-art loss functions, i.e., the cross entropy loss, focal loss, and true class probability confidence loss, are prone to view both correct predictions and incorrect predictions to be trustworthy. The reasons are two-fold. Firstly, correct predictions are generally dominant over incorrect predictions. Secondly, due to the data complexity, it is challenging to differentiate the incorrect predictions from the correct ones on real-world large-scale datasets. To improve the generalizability of trustworthiness predictors, we propose a novel steep slope loss to separate the features w.r.t. correct predictions from the ones w.r.t. incorrect predictions by two slide-like curves that oppose each other. The proposed loss is evaluated with two representative deep learning models, i.e., Vision Transformer and ResNet, as trustworthiness predictors. We conduct comprehensive experiments and analyses on ImageNet, which show that the proposed loss effectively improves the generalizability of trustworthiness predictors. The code and pre-trained trustworthiness predictors for reproducibility are available at https://github.com/luoyan407/predict_trustworthiness." @default.
• W3204195324 created "2021-10-11" @default.
• W3204195324 creator A5016415049 @default.
• W3204195324 creator A5020006712 @default.
• W3204195324 creator A5047419128 @default.
• W3204195324 creator A5055879357 @default.
• W3204195324 date "2021-09-30" @default.
• W3204195324 modified "2023-09-25" @default.
• W3204195324 title "Learning to Predict Trustworthiness with Steep Slope Loss" @default.
• W3204195324 cites W1503398984 @default.
• W3204195324 cites W1524761913 @default.
• W3204195324 cites W1580788756 @default.
• W3204195324 cites W1932198206 @default.
• W3204195324 cites W1944672 @default.
• W3204195324 cites W1965555277 @default.
• W3204195324 cites W1990824483 @default.
• W3204195324 cites W2108598243 @default.
• W3204195324 cites W2112796928 @default.
• W3204195324 cites W2121423746 @default.
• W3204195324 cites W2149298154 @default.
• W3204195324 cites W2163605009 @default.
• W3204195324 cites W2194775991 @default.
• W3204195324 cites W2202088841 @default.
• W3204195324 cites W2440599146 @default.
• W3204195324 cites W2531327146 @default.
• W3204195324 cites W2594522244 @default.
• W3204195324 cites W2902617128 @default.
• W3204195324 cites W2937223285 @default.
• W3204195324 cites W2955425717 @default.
• W3204195324 cites W2963060032 @default.
• W3204195324 cites W2963207607 @default.
• W3204195324 cites W2963351448 @default.
• W3204195324 cites W2963441669 @default.
• W3204195324 cites W2963613748 @default.
• W3204195324 cites W2963656735 @default.
• W3204195324 cites W2963691377 @default.
• W3204195324 cites W2964059111 @default.
• W3204195324 cites W2970390122 @default.
• W3204195324 cites W2970565181 @default.
• W3204195324 cites W2970941190 @default.
• W3204195324 cites W2970971581 @default.
• W3204195324 cites W3010512657 @default.
• W3204195324 cites W3034711780 @default.
• W3204195324 cites W3035059588 @default.
• W3204195324 cites W3099049390 @default.
• W3204195324 cites W3118608800 @default.
• W3204195324 cites W3119786062 @default.
• W3204195324 cites W608975221 @default.
• W3204195324 hasPublicationYear "2021" @default.
• W3204195324 type Work @default.
• W3204195324 sameAs 3204195324 @default.
• W3204195324 citedByCount "0" @default.
• W3204195324 crossrefType "posted-content" @default.
• W3204195324 hasAuthorship W3204195324A5016415049 @default.
• W3204195324 hasAuthorship W3204195324A5020006712 @default.
• W3204195324 hasAuthorship W3204195324A5047419128 @default.
• W3204195324 hasAuthorship W3204195324A5055879357 @default.
• W3204195324 hasBestOaLocation W32041953241 @default.
• W3204195324 hasConcept C105795698 @default.
• W3204195324 hasConcept C119857082 @default.
• W3204195324 hasConcept C121332964 @default.
• W3204195324 hasConcept C124101348 @default.
• W3204195324 hasConcept C153701036 @default.
• W3204195324 hasConcept C154945302 @default.
• W3204195324 hasConcept C17744445 @default.
• W3204195324 hasConcept C199539241 @default.
• W3204195324 hasConcept C27158222 @default.
• W3204195324 hasConcept C2778755073 @default.
• W3204195324 hasConcept C2780224610 @default.
• W3204195324 hasConcept C33923547 @default.
• W3204195324 hasConcept C38652104 @default.
• W3204195324 hasConcept C41008148 @default.
• W3204195324 hasConcept C62520636 @default.
• W3204195324 hasConcept C95623464 @default.
• W3204195324 hasConceptScore W3204195324C105795698 @default.
• W3204195324 hasConceptScore W3204195324C119857082 @default.
• W3204195324 hasConceptScore W3204195324C121332964 @default.
• W3204195324 hasConceptScore W3204195324C124101348 @default.
• W3204195324 hasConceptScore W3204195324C153701036 @default.
• W3204195324 hasConceptScore W3204195324C154945302 @default.
• W3204195324 hasConceptScore W3204195324C17744445 @default.
• W3204195324 hasConceptScore W3204195324C199539241 @default.
• W3204195324 hasConceptScore W3204195324C27158222 @default.
• W3204195324 hasConceptScore W3204195324C2778755073 @default.
• W3204195324 hasConceptScore W3204195324C2780224610 @default.
• W3204195324 hasConceptScore W3204195324C33923547 @default.
• W3204195324 hasConceptScore W3204195324C38652104 @default.
• W3204195324 hasConceptScore W3204195324C41008148 @default.
• W3204195324 hasConceptScore W3204195324C62520636 @default.
• W3204195324 hasConceptScore W3204195324C95623464 @default.
• W3204195324 hasLocation W32041953241 @default.
• W3204195324 hasOpenAccess W3204195324 @default.
• W3204195324 hasPrimaryLocation W32041953241 @default.
• W3204195324 hasRelatedWork W10719664 @default.
• W3204195324 hasRelatedWork W12829028 @default.
• W3204195324 hasRelatedWork W13034104 @default.
• W3204195324 hasRelatedWork W14115579 @default.
• W3204195324 hasRelatedWork W482721 @default.
• W3204195324 hasRelatedWork W6680660 @default.
• W3204195324 hasRelatedWork W728297 @default.

• next