FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3216234792> ?p ?o ?g. }

• W3216234792 abstract "Intuitively, one would expect accuracy of a trained neural network's prediction on test samples to correlate with how densely the samples are surrounded by seen training samples in representation space. We find that a bound on empirical training error smoothed across linear activation regions scales inversely with training sample density in representation space. Empirically, we verify this bound is a strong predictor of the inaccuracy of the network's prediction on test samples. For unseen test sets, including those with out-of-distribution samples, ranking test samples by their local region's error bound and discarding samples with the highest bounds raises prediction accuracy by up to 20% in absolute terms for image classification datasets, on average over thresholds." @default.
• W3216234792 created "2021-12-06" @default.
• W3216234792 creator A5012859570 @default.
• W3216234792 creator A5023545041 @default.
• W3216234792 creator A5043910056 @default.
• W3216234792 creator A5060511349 @default.
• W3216234792 creator A5062650781 @default.
• W3216234792 date "2021-06-15" @default.
• W3216234792 modified "2023-09-26" @default.
• W3216234792 title "Test Sample Accuracy Scales with Training Sample Density in Neural Networks" @default.
• W3216234792 cites W1580375566 @default.
• W3216234792 cites W1686946872 @default.
• W3216234792 cites W1944672 @default.
• W3216234792 cites W1995875735 @default.
• W3216234792 cites W2142635246 @default.
• W3216234792 cites W2147947791 @default.
• W3216234792 cites W2161388792 @default.
• W3216234792 cites W2169413195 @default.
• W3216234792 cites W2335728318 @default.
• W3216234792 cites W2531327146 @default.
• W3216234792 cites W2867167548 @default.
• W3216234792 cites W2963238274 @default.
• W3216234792 cites W2963441669 @default.
• W3216234792 cites W2963613748 @default.
• W3216234792 cites W2963693742 @default.
• W3216234792 cites W2963982496 @default.
• W3216234792 cites W2964059111 @default.
• W3216234792 cites W2970338090 @default.
• W3216234792 cites W2970641149 @default.
• W3216234792 cites W2970859221 @default.
• W3216234792 cites W2990500698 @default.
• W3216234792 cites W2996179709 @default.
• W3216234792 cites W3035898727 @default.
• W3216234792 cites W3104110041 @default.
• W3216234792 cites W3108981297 @default.
• W3216234792 cites W3115571254 @default.
• W3216234792 cites W3118608800 @default.
• W3216234792 cites W3125314187 @default.
• W3216234792 cites W3168441138 @default.
• W3216234792 doi "https://doi.org/10.48550/arxiv.2106.08365" @default.
• W3216234792 hasPublicationYear "2021" @default.
• W3216234792 type Work @default.
• W3216234792 sameAs 3216234792 @default.
• W3216234792 citedByCount "0" @default.
• W3216234792 crossrefType "posted-content" @default.
• W3216234792 hasAuthorship W3216234792A5012859570 @default.
• W3216234792 hasAuthorship W3216234792A5023545041 @default.
• W3216234792 hasAuthorship W3216234792A5043910056 @default.
• W3216234792 hasAuthorship W3216234792A5060511349 @default.
• W3216234792 hasAuthorship W3216234792A5062650781 @default.
• W3216234792 hasBestOaLocation W32162347921 @default.
• W3216234792 hasConcept C105795698 @default.
• W3216234792 hasConcept C111919701 @default.
• W3216234792 hasConcept C119857082 @default.
• W3216234792 hasConcept C129848803 @default.
• W3216234792 hasConcept C134306372 @default.
• W3216234792 hasConcept C151730666 @default.
• W3216234792 hasConcept C153180895 @default.
• W3216234792 hasConcept C153294291 @default.
• W3216234792 hasConcept C154945302 @default.
• W3216234792 hasConcept C17744445 @default.
• W3216234792 hasConcept C185592680 @default.
• W3216234792 hasConcept C189430467 @default.
• W3216234792 hasConcept C198531522 @default.
• W3216234792 hasConcept C199539241 @default.
• W3216234792 hasConcept C205649164 @default.
• W3216234792 hasConcept C2776359362 @default.
• W3216234792 hasConcept C2777211547 @default.
• W3216234792 hasConcept C2777267654 @default.
• W3216234792 hasConcept C2778572836 @default.
• W3216234792 hasConcept C33923547 @default.
• W3216234792 hasConcept C41008148 @default.
• W3216234792 hasConcept C43617362 @default.
• W3216234792 hasConcept C50644808 @default.
• W3216234792 hasConcept C77553402 @default.
• W3216234792 hasConcept C86803240 @default.
• W3216234792 hasConcept C94625758 @default.
• W3216234792 hasConceptScore W3216234792C105795698 @default.
• W3216234792 hasConceptScore W3216234792C111919701 @default.
• W3216234792 hasConceptScore W3216234792C119857082 @default.
• W3216234792 hasConceptScore W3216234792C129848803 @default.
• W3216234792 hasConceptScore W3216234792C134306372 @default.
• W3216234792 hasConceptScore W3216234792C151730666 @default.
• W3216234792 hasConceptScore W3216234792C153180895 @default.
• W3216234792 hasConceptScore W3216234792C153294291 @default.
• W3216234792 hasConceptScore W3216234792C154945302 @default.
• W3216234792 hasConceptScore W3216234792C17744445 @default.
• W3216234792 hasConceptScore W3216234792C185592680 @default.
• W3216234792 hasConceptScore W3216234792C189430467 @default.
• W3216234792 hasConceptScore W3216234792C198531522 @default.
• W3216234792 hasConceptScore W3216234792C199539241 @default.
• W3216234792 hasConceptScore W3216234792C205649164 @default.
• W3216234792 hasConceptScore W3216234792C2776359362 @default.
• W3216234792 hasConceptScore W3216234792C2777211547 @default.
• W3216234792 hasConceptScore W3216234792C2777267654 @default.
• W3216234792 hasConceptScore W3216234792C2778572836 @default.
• W3216234792 hasConceptScore W3216234792C33923547 @default.
• W3216234792 hasConceptScore W3216234792C41008148 @default.
• W3216234792 hasConceptScore W3216234792C43617362 @default.
• W3216234792 hasConceptScore W3216234792C50644808 @default.

• next