FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4360898052> ?p ?o ?g. }

• W4360898052 endingPage "5907" @default.
• W4360898052 startingPage "5894" @default.
• W4360898052 abstract "Generalization of the person re-identification (ReID) model plays an important role in practical application, and we discuss a simple yet effective regularizer to improve it inspired by Adversarial Training (AT). AT has been indicated as an advanced regularizer due to its adversarial mechanism, ability to mine hard samples, and nature of data augmentation. However, serving as an augmentation-based regularizer, AT shows low diversity of the perturbation, excessive computational cost, and the optimization dilemma between adversarial robustness and accuracy for ReID task, and is thus suboptimal. To tackle these limitations and get a more effective regularizer for ReID, we rethink the nature of AT and unveil that the adversarial data augmentation is essentially reflected by gradients. Based on this, a novel implicit regularizer, named Stochastic Gradient Perturbation (SGP), is proposed, which naturally brings three merits: 1) Better diversity of the perturbation due to the proposed non-directional stochastic perturbations rather than directional adversarial perturbations. 2) Lower computational cost due to the proposed implicit gradient augmentation rather than explicitly additional data. 3) The optimization dilemma of the adversarial robustness and generalization is naturally overcome since SGP contains the adversarial gradient perturbation. Further, we put forward a perspective that the generalization and adversarial robustness may have an inter unity. Experiments on the baseline and SOTA models demonstrate powerful performances of the plugged-played SGP, and both generalization and adversarial robustness can be guaranteed." @default.
• W4360898052 created "2023-03-26" @default.
• W4360898052 creator A5003381889 @default.
• W4360898052 creator A5028808297 @default.
• W4360898052 creator A5075091464 @default.
• W4360898052 creator A5076830444 @default.
• W4360898052 creator A5085955762 @default.
• W4360898052 date "2023-10-01" @default.
• W4360898052 modified "2023-10-05" @default.
• W4360898052 title "Stochastic Gradient Perturbation: An Implicit Regularizer for Person Re-Identification" @default.
• W4360898052 cites W1982925187 @default.
• W4360898052 cites W2104094955 @default.
• W4360898052 cites W2120587290 @default.
• W4360898052 cites W2180612164 @default.
• W4360898052 cites W2194775991 @default.
• W4360898052 cites W2204750386 @default.
• W4360898052 cites W2338117771 @default.
• W4360898052 cites W2585635281 @default.
• W4360898052 cites W2752695358 @default.
• W4360898052 cites W2768346313 @default.
• W4360898052 cites W2774644650 @default.
• W4360898052 cites W2777534232 @default.
• W4360898052 cites W2778652957 @default.
• W4360898052 cites W2904427185 @default.
• W4360898052 cites W2909593838 @default.
• W4360898052 cites W2912406707 @default.
• W4360898052 cites W2918380492 @default.
• W4360898052 cites W2947319827 @default.
• W4360898052 cites W2963000559 @default.
• W4360898052 cites W2963047834 @default.
• W4360898052 cites W2963383990 @default.
• W4360898052 cites W2963446712 @default.
• W4360898052 cites W2963842104 @default.
• W4360898052 cites W2963857521 @default.
• W4360898052 cites W2963975998 @default.
• W4360898052 cites W2963989829 @default.
• W4360898052 cites W2964097310 @default.
• W4360898052 cites W2967515867 @default.
• W4360898052 cites W2998126692 @default.
• W4360898052 cites W2998792609 @default.
• W4360898052 cites W3009153202 @default.
• W4360898052 cites W3016193692 @default.
• W4360898052 cites W3034273919 @default.
• W4360898052 cites W3035673257 @default.
• W4360898052 cites W3047363187 @default.
• W4360898052 cites W3047929456 @default.
• W4360898052 cites W3048070442 @default.
• W4360898052 cites W3093070413 @default.
• W4360898052 cites W3100506510 @default.
• W4360898052 cites W3101202235 @default.
• W4360898052 cites W3103557498 @default.
• W4360898052 cites W3108381568 @default.
• W4360898052 cites W3157802068 @default.
• W4360898052 cites W3166792421 @default.
• W4360898052 cites W3173909755 @default.
• W4360898052 cites W3185095134 @default.
• W4360898052 cites W3187541268 @default.
• W4360898052 cites W3191697454 @default.
• W4360898052 cites W3202788649 @default.
• W4360898052 cites W3203755984 @default.
• W4360898052 cites W3209159014 @default.
• W4360898052 cites W4200316912 @default.
• W4360898052 cites W4200343840 @default.
• W4360898052 cites W4210341392 @default.
• W4360898052 cites W4214736485 @default.
• W4360898052 cites W4226058100 @default.
• W4360898052 cites W4226265242 @default.
• W4360898052 cites W4285105056 @default.
• W4360898052 cites W4312311075 @default.
• W4360898052 cites W4312469934 @default.
• W4360898052 cites W4312622050 @default.
• W4360898052 cites W4313049560 @default.
• W4360898052 cites W4313069894 @default.
• W4360898052 cites W4313163489 @default.
• W4360898052 doi "https://doi.org/10.1109/tcsvt.2023.3261333" @default.
• W4360898052 hasPublicationYear "2023" @default.
• W4360898052 type Work @default.
• W4360898052 citedByCount "0" @default.
• W4360898052 crossrefType "journal-article" @default.
• W4360898052 hasAuthorship W4360898052A5003381889 @default.
• W4360898052 hasAuthorship W4360898052A5028808297 @default.
• W4360898052 hasAuthorship W4360898052A5075091464 @default.
• W4360898052 hasAuthorship W4360898052A5076830444 @default.
• W4360898052 hasAuthorship W4360898052A5085955762 @default.
• W4360898052 hasConcept C104317684 @default.
• W4360898052 hasConcept C11413529 @default.
• W4360898052 hasConcept C121332964 @default.
• W4360898052 hasConcept C126255220 @default.
• W4360898052 hasConcept C134306372 @default.
• W4360898052 hasConcept C154945302 @default.
• W4360898052 hasConcept C177148314 @default.
• W4360898052 hasConcept C177918212 @default.
• W4360898052 hasConcept C185592680 @default.
• W4360898052 hasConcept C33923547 @default.
• W4360898052 hasConcept C37736160 @default.
• W4360898052 hasConcept C41008148 @default.
• W4360898052 hasConcept C55493867 @default.
• W4360898052 hasConcept C62520636 @default.

• next