Zobrazeno 1 - 10
of 218
pro vyhledávání: '"Bors, Adrian"'
Deeper Vision Transformers (ViTs) are more challenging to train. We expose a degradation problem in deeper layers of ViT when using masked image modeling (MIM) for pre-training. To ease the training of deeper ViTs, we introduce a self-supervised lear
Externí odkaz:
http://arxiv.org/abs/2309.14136
Autor:
Huang, Guoxi, Bors, Adrian G.
Static appearance of video may impede the ability of a deep neural network to learn motion-relevant features in video action recognition. In this paper, we introduce a new concept, Dynamic Appearance (DA), summarizing the appearance information relat
Externí odkaz:
http://arxiv.org/abs/2211.12748
Autor:
Ye, Fei, Bors, Adrian G.
Learning from non-stationary data streams, also called Task-Free Continual Learning (TFCL) remains challenging due to the absence of explicit task information. Although recently some methods have been proposed for TFCL, they lack theoretical guarante
Externí odkaz:
http://arxiv.org/abs/2210.06579
Autor:
Ye, Fei, Bors, Adrian G.
Due to their inference, data representation and reconstruction properties, Variational Autoencoders (VAE) have been successfully used in continual learning classification tasks. However, their ability to generate images with specifications correspond
Externí odkaz:
http://arxiv.org/abs/2207.10131
Autor:
Ye, Fei, Bors, Adrian G.
Recently, continual learning (CL) has gained significant interest because it enables deep learning models to acquire new knowledge without forgetting previously learnt information. However, most existing works require knowing the task identities and
Externí odkaz:
http://arxiv.org/abs/2207.05080
Autor:
Ye, Fei, Bors, Adrian G.
In this article, we provide the appendix for Lifelong Generative Modelling Using Dynamic Expansion Graph Model. This appendix includes additional visual results as well as the numerical results on the challenging datasets. In addition, we also provid
Externí odkaz:
http://arxiv.org/abs/2203.13503
Autor:
Ye, Fei, Bors, Adrian G.
Variational Autoencoders (VAEs) suffer from degenerated performance, when learning several successive tasks. This is caused by catastrophic forgetting. In order to address the knowledge loss, VAEs are using either Generative Replay (GR) mechanisms or
Externí odkaz:
http://arxiv.org/abs/2112.08370
Autor:
Ye, Fei, Bors, Adrian G.
Recent research efforts in lifelong learning propose to grow a mixture of models to adapt to an increasing number of tasks. The proposed methodology shows promising results in overcoming catastrophic forgetting. However, the theory behind these succe
Externí odkaz:
http://arxiv.org/abs/2108.12278
Autor:
Ye, Fei, Bors, Adrian G.
Learning disentangled and interpretable representations is an important step towards accomplishing comprehensive data representations on the manifold. In this paper, we propose a novel representation learning algorithm which combines the inference ab
Externí odkaz:
http://arxiv.org/abs/2107.04705