Matches in SemOpenAlex for { <https://semopenalex.org/work/W2995813858> ?p ?o ?g. }
Showing items 1 to 85 of
85
with 100 items per page.
- W2995813858 abstract "This paper is aimed at creating extremely small and fast convolutional neural networks (CNN) for the problem of facial expression recognition (FER) from frontal face images. To this end, we employed the popular knowledge distillation (KD) method and identified two major shortcomings with its use: 1) a fine-grained grid search is needed for tuning the temperature hyperparameter and 2) to find the optimal size-accuracy balance, one needs to search for the final network size (or the compression rate). On the other hand, KD is proved to be useful for model compression for the FER problem, and we discovered that its effects get more and more significant with decreasing model size. In addition, we hypothesized that translation invariance achieved using max-pooling layers would not be useful for the FER problem as the expressions are sensitive to small, pixelwise changes around the eye and the mouth. However, we have found an intriguing improvement in generalization when max-pooling is used. We conducted experiments on two widelyused FER datasets, CK+ and Oulu-CASIA. Our smallest model (MicroExpNet), obtained using knowledge distillation, is less than 1MB in size and works at 1851 frames per second on an Intel i7 CPU. Despite being less accurate than the state-of-the-art, MicroExpNet still provides significant insights for designing a microarchitecture for the FER problem." @default.
- W2995813858 created "2019-12-26" @default.
- W2995813858 creator A5009474947 @default.
- W2995813858 creator A5010763397 @default.
- W2995813858 creator A5076398399 @default.
- W2995813858 date "2019-11-01" @default.
- W2995813858 modified "2023-09-23" @default.
- W2995813858 title "MicroExpNet: An Extremely Small and Fast Model For Expression Recognition From Face Images" @default.
- W2995813858 cites W132448360 @default.
- W2995813858 cites W1974210421 @default.
- W2995813858 cites W1996958556 @default.
- W2995813858 cites W2035372623 @default.
- W2995813858 cites W2039051707 @default.
- W2995813858 cites W2073534683 @default.
- W2995813858 cites W2097117768 @default.
- W2995813858 cites W2102570318 @default.
- W2995813858 cites W2103943262 @default.
- W2995813858 cites W2117539524 @default.
- W2995813858 cites W2141890865 @default.
- W2995813858 cites W2183341477 @default.
- W2995813858 cites W2217426128 @default.
- W2995813858 cites W2244142460 @default.
- W2995813858 cites W2251198138 @default.
- W2995813858 cites W2277498883 @default.
- W2995813858 cites W2490049321 @default.
- W2995813858 cites W2963112684 @default.
- W2995813858 cites W2963623198 @default.
- W2995813858 cites W3097096317 @default.
- W2995813858 cites W3100876745 @default.
- W2995813858 doi "https://doi.org/10.1109/ipta.2019.8936114" @default.
- W2995813858 hasPublicationYear "2019" @default.
- W2995813858 type Work @default.
- W2995813858 sameAs 2995813858 @default.
- W2995813858 citedByCount "24" @default.
- W2995813858 countsByYear W29958138582018 @default.
- W2995813858 countsByYear W29958138582019 @default.
- W2995813858 countsByYear W29958138582020 @default.
- W2995813858 countsByYear W29958138582021 @default.
- W2995813858 countsByYear W29958138582022 @default.
- W2995813858 countsByYear W29958138582023 @default.
- W2995813858 crossrefType "proceedings-article" @default.
- W2995813858 hasAuthorship W2995813858A5009474947 @default.
- W2995813858 hasAuthorship W2995813858A5010763397 @default.
- W2995813858 hasAuthorship W2995813858A5076398399 @default.
- W2995813858 hasBestOaLocation W29958138582 @default.
- W2995813858 hasConcept C144024400 @default.
- W2995813858 hasConcept C153180895 @default.
- W2995813858 hasConcept C154945302 @default.
- W2995813858 hasConcept C199360897 @default.
- W2995813858 hasConcept C2779304628 @default.
- W2995813858 hasConcept C2987714656 @default.
- W2995813858 hasConcept C31510193 @default.
- W2995813858 hasConcept C31972630 @default.
- W2995813858 hasConcept C36289849 @default.
- W2995813858 hasConcept C41008148 @default.
- W2995813858 hasConcept C90559484 @default.
- W2995813858 hasConceptScore W2995813858C144024400 @default.
- W2995813858 hasConceptScore W2995813858C153180895 @default.
- W2995813858 hasConceptScore W2995813858C154945302 @default.
- W2995813858 hasConceptScore W2995813858C199360897 @default.
- W2995813858 hasConceptScore W2995813858C2779304628 @default.
- W2995813858 hasConceptScore W2995813858C2987714656 @default.
- W2995813858 hasConceptScore W2995813858C31510193 @default.
- W2995813858 hasConceptScore W2995813858C31972630 @default.
- W2995813858 hasConceptScore W2995813858C36289849 @default.
- W2995813858 hasConceptScore W2995813858C41008148 @default.
- W2995813858 hasConceptScore W2995813858C90559484 @default.
- W2995813858 hasLocation W29958138581 @default.
- W2995813858 hasLocation W29958138582 @default.
- W2995813858 hasOpenAccess W2995813858 @default.
- W2995813858 hasPrimaryLocation W29958138581 @default.
- W2995813858 hasRelatedWork W1548715306 @default.
- W2995813858 hasRelatedWork W1560697087 @default.
- W2995813858 hasRelatedWork W1989039360 @default.
- W2995813858 hasRelatedWork W2060029454 @default.
- W2995813858 hasRelatedWork W2100085003 @default.
- W2995813858 hasRelatedWork W2136485282 @default.
- W2995813858 hasRelatedWork W2146295394 @default.
- W2995813858 hasRelatedWork W2347601237 @default.
- W2995813858 hasRelatedWork W2545171730 @default.
- W2995813858 hasRelatedWork W2908959303 @default.
- W2995813858 isParatext "false" @default.
- W2995813858 isRetracted "false" @default.
- W2995813858 magId "2995813858" @default.
- W2995813858 workType "article" @default.