2 citations · 2 across the 9 of their papers we have counts for
9 papers · 1 filter
Efficient Visual Representation Learning with Heat Conduction Equation
Zhemin Zhang, Xun Gong
Foundation models, such as CNNs and ViTs, have powered the development of image representation learning. However, general guidance to model architecture design is still missing. In…
Generating Multi-Center Classifier via Conditional Gaussian Distribution
Zhemin Zhang, Xun Gong
The linear classifier is widely used in various image classification tasks. It works by optimizing the distance between a sample and its corresponding class center. However, in rea…
Vision Big Bird: Random Sparsification for Full Attention
Zhemin Zhang, Xun Gong
Recently, Transformers have shown promising performance in various vision tasks. However, the high costs of global self-attention remain challenging for Transformers, especially fo…
RSIR Transformer: Hierarchical Vision Transformer using Random Sampling Windows and Important Region Windows
Zhemin Zhang, Xun Gong
Recently, Transformers have shown promising performance in various vision tasks. However, the high costs of global self-attention remain challenging for Transformers, especially fo…
Axially Expanded Windows for Local-Global Interaction in Vision Transformers
Zhemin Zhang, Xun Gong
Recently, Transformers have shown promising performance in various vision tasks. A challenging issue in Transformer design is that global self-attention is very expensive to comput…
Self-Supervised Implicit Attention: Guided Attention by The Model Itself
Jinyi Wu, Xun Gong, Zhemin Zhang
We propose Self-Supervised Implicit Attention (SSIA), a new approach that adaptively guides deep neural network models to gain attention by exploiting the properties of the models…
Positional Label for Self-Supervised Vision Transformer
Zhemin Zhang, Xun Gong
Positional encoding is important for vision transformer (ViT) to capture the spatial structure of the input image. General effectiveness has been proven in ViT. In our work we prop…
ReplaceBlock: An improved regularization method based on background information
Zhemin Zhang, Xun Gong, Jinyi Wu
Attention mechanism, being frequently used to train networks for better feature representations, can effectively disentangle the target object from irrelevant objects in the backgr…
The Fixed Sub-Center: A Better Way to Capture Data Complexity
Zhemin Zhang, Xun Gong
Treating class with a single center may hardly capture data distribution complexities. Using multiple sub-centers is an alternative way to address this problem. However, highly cor…