Zobrazeno 1 - 10
of 143
pro vyhledávání: '"Liu,Xialei"'
Class-incremental learning is a challenging problem, where the goal is to train a model that can classify data from an increasing number of classes over time. With the advancement of vision-language pre-trained models such as CLIP, they demonstrate g
Externí odkaz:
http://arxiv.org/abs/2407.14143
Class incremental semantic segmentation aims to preserve old knowledge while learning new tasks, however, it is impeded by catastrophic forgetting and background shift issues. Prior works indicate the pivotal importance of initializing new classifier
Externí odkaz:
http://arxiv.org/abs/2407.14142
In class-incremental learning (CIL) scenarios, the phenomenon of catastrophic forgetting caused by the classifier's bias towards the current task has long posed a significant challenge. It is mainly caused by the characteristic of discriminative mode
Externí odkaz:
http://arxiv.org/abs/2403.18383
Given unlabelled datasets containing both old and new categories, generalized category discovery (GCD) aims to accurately discover new classes while correctly classifying old classes, leveraging the class concepts learned from labeled samples. Curren
Externí odkaz:
http://arxiv.org/abs/2403.09974
Non-exemplar class incremental learning aims to learn both the new and old tasks without accessing any training data from the past. This strict restriction enlarges the difficulty of alleviating catastrophic forgetting since all techniques can only b
Externí odkaz:
http://arxiv.org/abs/2312.12722
With the advent of large-scale pre-trained models, interest in adapting and exploiting them for continual learning scenarios has grown. In this paper, we propose an approach to exploiting pre-trained vision-language models (e.g. CLIP) that enables fu
Externí odkaz:
http://arxiv.org/abs/2310.20348
Class Incremental Learning (CIL) aims to sequentially learn new classes while avoiding catastrophic forgetting of previous knowledge. We propose to use Masked Autoencoders (MAEs) as efficient learners for CIL. MAEs were originally designed to learn u
Externí odkaz:
http://arxiv.org/abs/2308.12510
Autor:
Jin, Xin, Xiao, Jia-Wen, Han, Ling-Hao, Guo, Chunle, Liu, Xialei, Li, Chongyi, Cheng, Ming-Ming
Explicit calibration-based methods have dominated RAW image denoising under extremely low-light environments. However, these methods are impeded by several critical limitations: a) the explicit calibration process is both labor- and time-intensive, b
Externí odkaz:
http://arxiv.org/abs/2308.03448
Autor:
Zeng, Guifang1 (AUTHOR) zenggf6@mail2.sysu.edu.cn, Liu, Xialei1 (AUTHOR), Zheng, Zhouying1 (AUTHOR), Zhao, Jiali1 (AUTHOR), Zhuo, Wenfeng1 (AUTHOR), Bai, Zirui1 (AUTHOR), Lin, En1 (AUTHOR), Cai, Shanglin1 (AUTHOR), Cai, Chaonong1 (AUTHOR), Li, Peiping1 (AUTHOR) lipp8@mail.sysu.edu.cn, Zou, Baojia1 (AUTHOR) zoubj6@mail.sysu.edu.cn, Li, Jian1 (AUTHOR) lijian5@mail.sysu.edu.cn
Publikováno v:
Lipids in Health & Disease. 12/27/2024, Vol. 23 Issue 1, p1-15. 15p.
Exemplar-free Class Incremental Learning (EFCIL) aims to sequentially learn tasks with access only to data from the current one. EFCIL is of interest because it mitigates concerns about privacy and long-term storage of data, while at the same time al
Externí odkaz:
http://arxiv.org/abs/2212.08251