2023-11-28 更新
White-Box Transformers via Sparse Rate Reduction: Compression Is All There Is?
Authors:Yaodong Yu, Sam Buchanan, Druv Pai, Tianzhe Chu, Ziyang Wu, Shengbang Tong, Hao Bai, Yuexiang Zhai, Benjamin D. Haeffele, Yi Ma
In this paper, we contend that a natural objective of representation learning is to compress and transform the distribution of the data, say sets of tokens, towards a low-dimensional Gaussian mixture supported on incoherent subspaces. The goodness of such a representation can be evaluated by a principled measure, called sparse rate reduction, that simultaneously maximizes the intrinsic information gain and extrinsic sparsity of the learned representation. From this perspective, popular deep network architectures, including transformers, can be viewed as realizing iterative schemes to optimize this measure. Particularly, we derive a transformer block from alternating optimization on parts of this objective: the multi-head self-attention operator compresses the representation by implementing an approximate gradient descent step on the coding rate of the features, and the subsequent multi-layer perceptron sparsifies the features. This leads to a family of white-box transformer-like deep network architectures, named CRATE, which are mathematically fully interpretable. We show, by way of a novel connection between denoising and compression, that the inverse to the aforementioned compressive encoding can be realized by the same class of CRATE architectures. Thus, the so-derived white-box architectures are universal to both encoders and decoders. Experiments show that these networks, despite their simplicity, indeed learn to compress and sparsify representations of large-scale real-world image and text datasets, and achieve performance very close to highly engineered transformer-based models: ViT, MAE, DINO, BERT, and GPT2. We believe the proposed computational framework demonstrates great potential in bridging the gap between theory and practice of deep learning, from a unified perspective of data compression. Code is available at: https://ma-lab-berkeley.github.io/CRATE .
PDF This paper integrates the works arXiv:2306.01129 and arXiv:2308.16271 into a complete story. In this paper, we improve the writing and organization, and also add conceptual, empirical, and theoretical improvements over the previous work. V2: small typo fixes and formatting improvements
点此查看论文截图
Compositional Zero-shot Learning via Progressive Language-based Observations
Authors:Lin Li, Guikun Chen, Jun Xiao, Long Chen
Compositional zero-shot learning aims to recognize unseen state-object compositions by leveraging known primitives (state and object) during training. However, effectively modeling interactions between primitives and generalizing knowledge to novel compositions remains a perennial challenge. There are two key factors: object-conditioned and state-conditioned variance, i.e., the appearance of states (or objects) can vary significantly when combined with different objects (or states). For instance, the state “old” can signify a vintage design for a “car” or an advanced age for a “cat”. In this paper, we argue that these variances can be mitigated by predicting composition categories based on pre-observed primitive. To this end, we propose Progressive Language-based Observations (PLO), which can dynamically determine a better observation order of primitives. These observations comprise a series of concepts or languages that allow the model to understand image content in a step-by-step manner. Specifically, PLO adopts pre-trained vision-language models (VLMs) to empower the model with observation capabilities. We further devise two variants: 1) PLO-VLM: a two-step method, where a pre-observing classifier dynamically determines the observation order of two primitives. 2) PLO-LLM: a multi-step scheme, which utilizes large language models (LLMs) to craft composition-specific prompts for step-by-step observing. Extensive ablations on three challenging datasets demonstrate the superiority of PLO compared with state-of-the-art methods, affirming its abilities in compositional recognition.
PDF
点此查看论文截图
Advancing Vision Transformers with Group-Mix Attention
Authors:Chongjian Ge, Xiaohan Ding, Zhan Tong, Li Yuan, Jiangliu Wang, Yibing Song, Ping Luo
Vision Transformers (ViTs) have been shown to enhance visual recognition through modeling long-range dependencies with multi-head self-attention (MHSA), which is typically formulated as Query-Key-Value computation. However, the attention map generated from the Query and Key captures only token-to-token correlations at one single granularity. In this paper, we argue that self-attention should have a more comprehensive mechanism to capture correlations among tokens and groups (i.e., multiple adjacent tokens) for higher representational capacity. Thereby, we propose Group-Mix Attention (GMA) as an advanced replacement for traditional self-attention, which can simultaneously capture token-to-token, token-to-group, and group-to-group correlations with various group sizes. To this end, GMA splits the Query, Key, and Value into segments uniformly and performs different group aggregations to generate group proxies. The attention map is computed based on the mixtures of tokens and group proxies and used to re-combine the tokens and groups in Value. Based on GMA, we introduce a powerful backbone, namely GroupMixFormer, which achieves state-of-the-art performance in image classification, object detection, and semantic segmentation with fewer parameters than existing models. For instance, GroupMixFormer-L (with 70.3M parameters and 384^2 input) attains 86.2% Top-1 accuracy on ImageNet-1K without external data, while GroupMixFormer-B (with 45.8M parameters) attains 51.2% mIoU on ADE20K.
PDF
点此查看论文截图
ChAda-ViT : Channel Adaptive Attention for Joint Representation Learning of Heterogeneous Microscopy Images
Authors:Nicolas Bourriez, Ihab Bendidi, Ethan Cohen, Gabriel Watkinson, Maxime Sanchez, Guillaume Bollot, Auguste Genovesio
Unlike color photography images, which are consistently encoded into RGB channels, biological images encompass various modalities, where the type of microscopy and the meaning of each channel varies with each experiment. Importantly, the number of channels can range from one to a dozen and their correlation is often comparatively much lower than RGB, as each of them brings specific information content. This aspect is largely overlooked by methods designed out of the bioimage field, and current solutions mostly focus on intra-channel spatial attention, often ignoring the relationship between channels, yet crucial in most biological applications. Importantly, the variable channel type and count prevent the projection of several experiments to a unified representation for large scale pre-training. In this study, we propose ChAda-ViT, a novel Channel Adaptive Vision Transformer architecture employing an Inter-Channel Attention mechanism on images with an arbitrary number, order and type of channels. We also introduce IDRCell100k, a bioimage dataset with a rich set of 79 experiments covering 7 microscope modalities, with a multitude of channel types, and channel counts varying from 1 to 10 per experiment. Our proposed architecture, trained in a self-supervised manner, outperforms existing approaches in several biologically relevant downstream tasks. Additionally, it can be used to bridge the gap for the first time between assays with different microscopes, channel numbers or types by embedding various image and experimental modalities into a unified biological image representation. The latter should facilitate interdisciplinary studies and pave the way for better adoption of deep learning in biological image-based analyses. Code and Data to be released soon.
PDF
点此查看论文截图
Improving Adaptability and Generalizability of Efficient Transfer Learning for Vision-Language Models
Authors:Yongjin Yang, Jongwoo Ko, Se-Young Yun
Vision-Language Models (VLMs) like CLIP have demonstrated remarkable applicability across a variety of downstream tasks, including zero-shot image classification. Recently, the use of prompts or adapters for efficient transfer learning has gained significant attention for effectively adapting to downstream tasks. However, the roles of vision and text prompts, as well as adapters in terms of generalization and transfer difficulty, have been overlooked, limiting performance on unseen tasks. In this paper, we empirically analyze how VLMs behave when using vision and text prompts, adapters, and a combination of these components, marking a novel exploration by our study. Our observations find that utilizing vision prompts for class separability and text adapters for task adaptation is crucial for adaptability and generalizability. Moreover, to improve generalization across every domain, we propose an adaptive ensemble method that effectively combines the general knowledge of VLMs with task-specific knowledge according to transfer difficulty. Upon experimenting with extensive benchmarks, our method consistently outperforms all baselines, particularly on unseen tasks, demonstrating the effectiveness of our proposed approach.
PDF 11 pages (19 pages including supplementary), 10 figures (12 figures including supplementary), 6 tables (17 tables including supplementary)
点此查看论文截图
C-SAW: Self-Supervised Prompt Learning for Image Generalization in Remote Sensing
Authors:Avigyan Bhattacharya, Mainak Singha, Ankit Jha, Biplab Banerjee
We focus on domain and class generalization problems in analyzing optical remote sensing images, using the large-scale pre-trained vision-language model (VLM), CLIP. While contrastively trained VLMs show impressive zero-shot generalization performance, their effectiveness is limited when dealing with diverse domains during training and testing. Existing prompt learning techniques overlook the importance of incorporating domain and content information into the prompts, which results in a drop in performance while dealing with such multi-domain data. To address these challenges, we propose a solution that ensures domain-invariant prompt learning while enhancing the expressiveness of visual features. We observe that CLIP’s vision encoder struggles to identify contextual image information, particularly when image patches are jumbled up. This issue is especially severe in optical remote sensing images, where land-cover classes exhibit well-defined contextual appearances. To this end, we introduce C-SAW, a method that complements CLIP with a self-supervised loss in the visual space and a novel prompt learning technique that emphasizes both visual domain and content-specific features. We keep the CLIP backbone frozen and introduce a small set of projectors for both the CLIP encoders to train C-SAW contrastively. Experimental results demonstrate the superiority of C-SAW across multiple remote sensing benchmarks and different generalization tasks.
PDF Accepted in ACM ICVGIP 2023
点此查看论文截图
ViT-Lens-2: Gateway to Omni-modal Intelligence
Authors:Weixian Lei, Yixiao Ge, Kun Yi, Jianfeng Zhang, Difei Gao, Dylan Sun, Yuying Ge, Ying Shan, Mike Zheng Shou
Aiming to advance AI agents, large foundation models significantly improve reasoning and instruction execution, yet the current focus on vision and language neglects the potential of perceiving diverse modalities in open-world environments. However, the success of data-driven vision and language models is costly or even infeasible to be reproduced for rare modalities. In this paper, we present ViT-Lens-2 that facilitates efficient omni-modal representation learning by perceiving novel modalities with a pretrained ViT and aligning them to a pre-defined space. Specifically, the modality-specific lens is tuned to project any-modal signals to an intermediate embedding space, which are then processed by a strong ViT with pre-trained visual knowledge. The encoded representations are optimized toward aligning with the modal-independent space, pre-defined by off-the-shelf foundation models. ViT-Lens-2 provides a unified solution for representation learning of increasing modalities with two appealing advantages: (i) Unlocking the great potential of pretrained ViTs to novel modalities effectively with efficient data regime; (ii) Enabling emergent downstream capabilities through modality alignment and shared ViT parameters. We tailor ViT-Lens-2 to learn representations for 3D point cloud, depth, audio, tactile and EEG, and set new state-of-the-art results across various understanding tasks, such as zero-shot classification. By seamlessly integrating ViT-Lens-2 into Multimodal Foundation Models, we enable Any-modality to Text and Image Generation in a zero-shot manner. Code and models are available at https://github.com/TencentARC/ViT-Lens.
PDF This work is a follow-up of “ViT-Lens: Towards Omni-modal Representations”. arXiv admin note: text overlap with arXiv:2308.10185
点此查看论文截图