Towards Better Multi-head Attention via Channel-wise Sample Permutation

• URL: http://arxiv.org/abs/2410.10914v1
• Date: Mon, 14 Oct 2024 06:28:40 GMT
• Title: Towards Better Multi-head Attention via Channel-wise Sample Permutation
• Authors: Shen Yuan, Hongteng Xu,
• Abstract summary: Transformer plays a central role in many fundamental deep learning models, e.g., the ViT in computer vision and the BERT and GPT in natural language processing. We propose a simple and novel channel-wise sample permutation (CSP) operator, achieving a new structured MHA with fewer parameters and lower complexity.
• Score: 27.36959595444929
• License:
• Abstract: Transformer plays a central role in many fundamental deep learning models, e.g., the ViT in computer vision and the BERT and GPT in natural language processing, whose effectiveness is mainly attributed to its multi-head attention (MHA) mechanism. In this study, we propose a simple and novel channel-wise sample permutation (CSP) operator, achieving a new structured MHA with fewer parameters and lower complexity. Given an input matrix, CSP circularly shifts the samples of different channels with various steps and then sorts grouped samples of each channel. This operator is equivalent to implicitly implementing cross-channel attention maps as permutation matrices, which achieves linear complexity and suppresses the risk of rank collapse when representing data. We replace the MHA of some representative models with CSP and test the CSP-based models in several discriminative tasks, including image classification and long sequence analysis. Experiments show that the CSP-based models achieve comparable or better performance with fewer parameters and lower computational costs than the classic Transformer and its state-of-the-art variants. The code is available at https://github.com/DaShenZi721/CSP.

Related papers

• SCHEME: Scalable Channel Mixer for Vision Transformers [52.605868919281086]
  Vision Transformers have achieved impressive performance in many vision tasks. Much less research has been devoted to the channel mixer or feature mixing block (FFN or) We show that the dense connections can be replaced with a diagonal block structure that supports larger expansion ratios.
  arXiv  Detail & Related papers  (2023-12-01T08:22:34Z)
• Sliceformer: Make Multi-head Attention as Simple as Sorting in Discriminative Tasks [32.33355192614434]
  We propose an effective and efficient surrogate of the Transformer, called Sliceformer. Our Sliceformer replaces the classic MHA mechanism with an extremely simple slicing-sorting'' operation. Our Sliceformer achieves comparable or better performance with lower memory cost and faster speed than the Transformer and its variants.
  arXiv  Detail & Related papers  (2023-10-26T14:43:07Z)
• ClaSP -- Parameter-free Time Series Segmentation [6.533695062182296]
  We present ClaSP, a novel, highly accurate and domain-agnostic method for time series segmentation. ClaSP hierarchically splits a TS into two parts. A change point is determined by training a binary TS classifier for each possible split point and selecting the one split that is best at identifying subsequences to be from either of the partitions. In our experimental evaluation using a benchmark of 115 data sets, we show that ClaSP outperforms the state of the art in terms of accuracy and is fast and scalable.
  arXiv  Detail & Related papers  (2022-07-28T10:05:53Z)
• Multimodal Fusion Transformer for Remote Sensing Image Classification [35.57881383390397]
  Vision transformers (ViTs) have been trending in image classification tasks due to their promising performance when compared to convolutional neural networks (CNNs) To achieve satisfactory performance, close to that of CNNs, transformers need fewer parameters. We introduce a new multimodal fusion transformer (MFT) network which comprises a multihead cross patch attention (mCrossPA) for HSI land-cover classification.
  arXiv  Detail & Related papers  (2022-03-31T11:18:41Z)
• Rethinking Semantic Segmentation: A Prototype View [126.59244185849838]
  We present a nonparametric semantic segmentation model based on non-learnable prototypes. Our framework yields compelling results over several datasets. We expect this work will provoke a rethink of the current de facto semantic segmentation model design.
  arXiv  Detail & Related papers  (2022-03-28T21:15:32Z)
• Multi-scale and Cross-scale Contrastive Learning for Semantic Segmentation [5.281694565226513]
  We apply contrastive learning to enhance the discriminative power of the multi-scale features extracted by semantic segmentation networks. By first mapping the encoder's multi-scale representations to a common feature space, we instantiate a novel form of supervised local-global constraint.
  arXiv  Detail & Related papers  (2022-03-25T01:24:24Z)
• Dynamically-Scaled Deep Canonical Correlation Analysis [77.34726150561087]
  Canonical Correlation Analysis (CCA) is a method for feature extraction of two views by finding maximally correlated linear projections of them. We introduce a novel dynamic scaling method for training an input-dependent canonical correlation model.
  arXiv  Detail & Related papers  (2022-03-23T12:52:49Z)
• A new perspective on probabilistic image modeling [92.89846887298852]
  We present a new probabilistic approach for image modeling capable of density estimation, sampling and tractable inference. DCGMMs can be trained end-to-end by SGD from random initial conditions, much like CNNs. We show that DCGMMs compare favorably to several recent PC and SPN models in terms of inference, classification and sampling.
  arXiv  Detail & Related papers  (2022-03-21T14:53:57Z)
• Coarse-to-Fine Sparse Transformer for Hyperspectral Image Reconstruction [138.04956118993934]
  We propose a novel Transformer-based method, coarse-to-fine sparse Transformer (CST) CST embedding HSI sparsity into deep learning for HSI reconstruction. In particular, CST uses our proposed spectra-aware screening mechanism (SASM) for coarse patch selecting. Then the selected patches are fed into our customized spectra-aggregation hashing multi-head self-attention (SAH-MSA) for fine pixel clustering and self-similarity capturing.
  arXiv  Detail & Related papers  (2022-03-09T16:17:47Z)
• Kernel and Rich Regimes in Overparametrized Models [69.40899443842443]
  We show that gradient descent on overparametrized multilayer networks can induce rich implicit biases that are not RKHS norms. We also demonstrate this transition empirically for more complex matrix factorization models and multilayer non-linear networks.
  arXiv  Detail & Related papers  (2020-02-20T15:43:02Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.