DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Kweon, In So | - |
dc.contributor.advisor | 권인소 | - |
dc.contributor.author | Hur, Sungsu | - |
dc.date.accessioned | 2023-06-26T19:33:32Z | - |
dc.date.available | 2023-06-26T19:33:32Z | - |
dc.date.issued | 2023 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=1032947&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/309813 | - |
dc.description | 학위논문(석사) - 한국과학기술원 : 전기및전자공학부, 2023.2,[iv, 29 p. :] | - |
dc.description.abstract | Universal Domain Adaptation aims to transfer the knowledge between the datasets by handling two shifts: domain-shift and category-shift. The main challenge is correctly distinguishing the unknown target samples while adapting the distribution of known class knowledge from source to target. Most existing methods approach this problem by first training the target-adapted known classifier and then relying on the single threshold to distinguish unknown target samples. However, this simple threshold-based approach prevents the model from considering the underlying complexities existing between the known and unknown samples in the high-dimensional feature space. In this paper, we propose a new approach in which we use two sets of feature points, namely dual \textbf{C}lassifiers for \textbf{P}rototypes and \textbf{R}eciprocals (\textbf{CPR}). Our key idea is to associate each prototype with corresponding known class features while pushing the reciprocals apart from these prototypes to locate them in the potential unknown feature space. The target samples are then classified as unknown if they fall near any reciprocals at test time. To successfully train our framework, we collect the partial, confident target samples that are classified as known or unknown through our proposed multi-criteria selection. We then additionally apply the entropy loss regularization to them. For further adaptation, we also apply standard consistency regularization that matches the predictions of two different views of the input to make a more compact target feature space. We evaluate our proposal, CPR, on three standard benchmarks and achieve comparable or new state-of-the-art results. We also provide extensive ablation experiments to verify our main design choices in our framework. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | Deep learning▼adomain adaptation▼auniversal domain adaptation▼acategory shift▼areciprocal point | - |
dc.subject | 딥러닝▼a영역 적응▼a보편적 영역 적응▼a카테고리 변환▼a반대점 | - |
dc.title | Learning classifiers of prototypes and reciprocal points for universal domain adaptation | - |
dc.title.alternative | 임시 중점과 반대점 학습을 통한 보편적 영역 적응 | - |
dc.type | Thesis(Master) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :전기및전자공학부, | - |
dc.contributor.alternativeauthor | 허성수 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.