Transformer Fusion with Optimal Transport

• URL: http://arxiv.org/abs/2310.05719v3
• Date: Mon, 22 Apr 2024 09:38:17 GMT
• Title: Transformer Fusion with Optimal Transport
• Authors: Moritz Imfeld, Jacopo Graldi, Marco Giordano, Thomas Hofmann, Sotiris Anagnostidis, Sidak Pal Singh,
• Abstract summary: Fusion is a technique for merging multiple independently-trained neural networks in order to combine their capabilities. This paper presents a systematic approach for fusing two or more transformer-based networks exploiting Optimal Transport to (soft-)align the various architectural components.
• Score: 25.022849817421964
• License: http://creativecommons.org/licenses/by-sa/4.0/
• Abstract: Fusion is a technique for merging multiple independently-trained neural networks in order to combine their capabilities. Past attempts have been restricted to the case of fully-connected, convolutional, and residual networks. This paper presents a systematic approach for fusing two or more transformer-based networks exploiting Optimal Transport to (soft-)align the various architectural components. We flesh out an abstraction for layer alignment, that can generalize to arbitrary architectures - in principle - and we apply this to the key ingredients of Transformers such as multi-head self-attention, layer-normalization, and residual connections, and we discuss how to handle them via various ablation studies. Furthermore, our method allows the fusion of models of different sizes (heterogeneous fusion), providing a new and efficient way to compress Transformers. The proposed approach is evaluated on both image classification tasks via Vision Transformer and natural language modeling tasks using BERT. Our approach consistently outperforms vanilla fusion, and, after a surprisingly short finetuning, also outperforms the individual converged parent models. In our analysis, we uncover intriguing insights about the significant role of soft alignment in the case of Transformers. Our results showcase the potential of fusing multiple Transformers, thus compounding their expertise, in the budding paradigm of model fusion and recombination. Code is available at https://github.com/graldij/transformer-fusion.

Related papers

• Why mamba is effective? Exploit Linear Transformer-Mamba Network for Multi-Modality Image Fusion [15.79138560700532]
  We propose a dual-branch image fusion network called Tmamba. It consists of linear Transformer and Mamba, which has global modeling capabilities while maintaining linear complexity. Experiments show that our Tmamba achieves promising results in multiple fusion tasks, including infrared-visible image fusion and medical image fusion.
  arXiv  Detail & Related papers  (2024-09-05T03:42:11Z)
• CTRL-F: Pairing Convolution with Transformer for Image Classification via Multi-Level Feature Cross-Attention and Representation Learning Fusion [0.0]
  We present a novel lightweight hybrid network that pairs Convolution with Transformers. We fuse the local responses acquired from the convolution path with the global responses acquired from the MFCA module. Experiments demonstrate that our variants achieve state-of-the-art performance, whether trained from scratch on large data or even with low-data regime.
  arXiv  Detail & Related papers  (2024-07-09T08:47:13Z)
• Isomer: Isomerous Transformer for Zero-shot Video Object Segmentation [59.91357714415056]
  We propose two Transformer variants: Context-Sharing Transformer (CST) and Semantic Gathering-Scattering Transformer (S GST) CST learns the global-shared contextual information within image frames with a lightweight computation; S GST models the semantic correlation separately for the foreground and background. Compared with the baseline that uses vanilla Transformers for multi-stage fusion, ours significantly increase the speed by 13 times and achieves new state-of-the-art ZVOS performance.
  arXiv  Detail & Related papers  (2023-08-13T06:12:00Z)
• Multimodal Image Fusion based on Hybrid CNN-Transformer and Non-local Cross-modal Attention [12.167049432063132]
  We present a hybrid model consisting of a convolutional encoder and a Transformer-based decoder to fuse multimodal images. A branch fusion module is designed to adaptively fuse the features of the two branches.
  arXiv  Detail & Related papers  (2022-10-18T13:30:52Z)
• TransVG++: End-to-End Visual Grounding with Language Conditioned Vision Transformer [188.00681648113223]
  We explore neat yet effective Transformer-based frameworks for visual grounding. TransVG establishes multi-modal correspondences by Transformers and localizes referred regions by directly regressing box coordinates. We upgrade our framework to a purely Transformer-based one by leveraging Vision Transformer (ViT) for vision feature encoding.
  arXiv  Detail & Related papers  (2022-06-14T06:27:38Z)
• Adaptive Split-Fusion Transformer [90.04885335911729]
  We propose an Adaptive Split-Fusion Transformer (ASF-former) to treat convolutional and attention branches differently with adaptive weights. Experiments on standard benchmarks, such as ImageNet-1K, show that our ASF-former outperforms its CNN, transformer counterparts, and hybrid pilots in terms of accuracy.
  arXiv  Detail & Related papers  (2022-04-26T10:00:28Z)
• Multimodal Token Fusion for Vision Transformers [54.81107795090239]
  We propose a multimodal token fusion method (TokenFusion) for transformer-based vision tasks. To effectively fuse multiple modalities, TokenFusion dynamically detects uninformative tokens and substitutes these tokens with projected and aggregated inter-modal features. The design of TokenFusion allows the transformer to learn correlations among multimodal features, while the single-modal transformer architecture remains largely intact.
  arXiv  Detail & Related papers  (2022-04-19T07:47:50Z)
• Image Fusion Transformer [75.71025138448287]
  In image fusion, images obtained from different sensors are fused to generate a single image with enhanced information. In recent years, state-of-the-art methods have adopted Convolution Neural Networks (CNNs) to encode meaningful features for image fusion. We propose a novel Image Fusion Transformer (IFT) where we develop a transformer-based multi-scale fusion strategy.
  arXiv  Detail & Related papers  (2021-07-19T16:42:49Z)
• Parameter Efficient Multimodal Transformers for Video Representation Learning [108.8517364784009]
  This work focuses on reducing the parameters of multimodal Transformers in the context of audio-visual video representation learning. We show that our approach reduces parameters up to 80$%$, allowing us to train our model end-to-end from scratch. To demonstrate our approach, we pretrain our model on 30-second clips from Kinetics-700 and transfer it to audio-visual classification tasks.
  arXiv  Detail & Related papers  (2020-12-08T00:16:13Z)

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.