Matches in SemOpenAlex for { <https://semopenalex.org/work/W4385196562> ?p ?o ?g. }
- W4385196562 endingPage "18" @default.
- W4385196562 startingPage "1" @default.
- W4385196562 abstract "Recent years have witnessed a remarkable breakthrough in Synthetic Aperture Radar Automatic Target Recognition (SAR ATR) with the development of deep learning (DL). Nonetheless, once deployed, the DL-based methods’ ability to incrementally learn new knowledge from few-shot samples without forgetting the old is fragile, hindering them from discriminating unseen targets in real-world situations. In this paper, we propose a Cosine Prototype Learning (CPL) framework to first unlock few-shot class-incremental learning (FSCIL) in the SAR ATR field inspired by the intrinsic relationships between target azimuth-aware knowledge and semantic features under the cosine criterion. By condensing class-specific characteristics into individual prototypes, stable profiles of targets are depicted without losing generalization. For the model’s plasticity, a pairwise structure separation (PSS) loss is introduced to separate old and new classes and compact intra-class features. Meanwhile, the model’s transferability on new classes is guaranteed by a prototype consistency (PC) loss. For the model’s stability, we propose a prototype-exemplar distillation (PED) loss and a prototype re-calibration (PR) strategy to penalize semantic drifts of old-class feature spaces and alleviate the misalignment of the learned prototypes successively. At inference, a nearest-class-mean (NCM) classifier is adopted for evaluation by comparing cosine similarity scores between testing samples and class-specific prototypes. In experiments, the proposed components of our method are explored by ablation studies. Strong baselines are established, and extensive experiments conducted on the MSTAR dataset show that our method outperforms state-of-the-art methods under various FSCIL conditions, verifying its effectiveness for the FSCIL of SAR ATR." @default.
- W4385196562 created "2023-07-25" @default.
- W4385196562 creator A5004069573 @default.
- W4385196562 creator A5009225924 @default.
- W4385196562 creator A5021639465 @default.
- W4385196562 creator A5024965777 @default.
- W4385196562 creator A5071074935 @default.
- W4385196562 creator A5082887216 @default.
- W4385196562 date "2023-01-01" @default.
- W4385196562 modified "2023-10-13" @default.
- W4385196562 title "Few-Shot Class-Incremental SAR Target Recognition via Cosine Prototype Learning" @default.
- W4385196562 cites W1965572316 @default.
- W4385196562 cites W1998943499 @default.
- W4385196562 cites W2021739315 @default.
- W4385196562 cites W2038501427 @default.
- W4385196562 cites W2091636800 @default.
- W4385196562 cites W2106749053 @default.
- W4385196562 cites W2108598243 @default.
- W4385196562 cites W2111847198 @default.
- W4385196562 cites W2120066054 @default.
- W4385196562 cites W2147228188 @default.
- W4385196562 cites W2149918662 @default.
- W4385196562 cites W2160445017 @default.
- W4385196562 cites W2172078172 @default.
- W4385196562 cites W2194775991 @default.
- W4385196562 cites W2410591237 @default.
- W4385196562 cites W2615263668 @default.
- W4385196562 cites W2765793020 @default.
- W4385196562 cites W2883279772 @default.
- W4385196562 cites W2948594896 @default.
- W4385196562 cites W2948734064 @default.
- W4385196562 cites W2961699889 @default.
- W4385196562 cites W2962858109 @default.
- W4385196562 cites W2964105864 @default.
- W4385196562 cites W2964189064 @default.
- W4385196562 cites W2988916019 @default.
- W4385196562 cites W2990641522 @default.
- W4385196562 cites W3034942609 @default.
- W4385196562 cites W3035342403 @default.
- W4385196562 cites W3036195527 @default.
- W4385196562 cites W3038113894 @default.
- W4385196562 cites W3044680161 @default.
- W4385196562 cites W3099206234 @default.
- W4385196562 cites W3113584014 @default.
- W4385196562 cites W3169024950 @default.
- W4385196562 cites W3174868646 @default.
- W4385196562 cites W3174917907 @default.
- W4385196562 cites W3175270254 @default.
- W4385196562 cites W3175771944 @default.
- W4385196562 cites W3176779063 @default.
- W4385196562 cites W3177494822 @default.
- W4385196562 cites W3204291495 @default.
- W4385196562 cites W3216412612 @default.
- W4385196562 cites W4205385989 @default.
- W4385196562 cites W4212857616 @default.
- W4385196562 cites W4226134030 @default.
- W4385196562 cites W4296425607 @default.
- W4385196562 cites W4296704881 @default.
- W4385196562 doi "https://doi.org/10.1109/tgrs.2023.3298016" @default.
- W4385196562 hasPublicationYear "2023" @default.
- W4385196562 type Work @default.
- W4385196562 citedByCount "0" @default.
- W4385196562 crossrefType "journal-article" @default.
- W4385196562 hasAuthorship W4385196562A5004069573 @default.
- W4385196562 hasAuthorship W4385196562A5009225924 @default.
- W4385196562 hasAuthorship W4385196562A5021639465 @default.
- W4385196562 hasAuthorship W4385196562A5024965777 @default.
- W4385196562 hasAuthorship W4385196562A5071074935 @default.
- W4385196562 hasAuthorship W4385196562A5082887216 @default.
- W4385196562 hasConcept C117623542 @default.
- W4385196562 hasConcept C119857082 @default.
- W4385196562 hasConcept C153180895 @default.
- W4385196562 hasConcept C154945302 @default.
- W4385196562 hasConcept C2780762811 @default.
- W4385196562 hasConcept C41008148 @default.
- W4385196562 hasConcept C52622490 @default.
- W4385196562 hasConcept C87360688 @default.
- W4385196562 hasConcept C95623464 @default.
- W4385196562 hasConceptScore W4385196562C117623542 @default.
- W4385196562 hasConceptScore W4385196562C119857082 @default.
- W4385196562 hasConceptScore W4385196562C153180895 @default.
- W4385196562 hasConceptScore W4385196562C154945302 @default.
- W4385196562 hasConceptScore W4385196562C2780762811 @default.
- W4385196562 hasConceptScore W4385196562C41008148 @default.
- W4385196562 hasConceptScore W4385196562C52622490 @default.
- W4385196562 hasConceptScore W4385196562C87360688 @default.
- W4385196562 hasConceptScore W4385196562C95623464 @default.
- W4385196562 hasFunder F4320321001 @default.
- W4385196562 hasFunder F4320335777 @default.
- W4385196562 hasLocation W43851965621 @default.
- W4385196562 hasOpenAccess W4385196562 @default.
- W4385196562 hasPrimaryLocation W43851965621 @default.
- W4385196562 hasRelatedWork W1964120219 @default.
- W4385196562 hasRelatedWork W2062264607 @default.
- W4385196562 hasRelatedWork W2085350532 @default.
- W4385196562 hasRelatedWork W2115092356 @default.
- W4385196562 hasRelatedWork W2146076056 @default.
- W4385196562 hasRelatedWork W2385132419 @default.