arXiv 2206.058102022-06-12cs.CVcs.LGcs.NE

Analysis of Branch Specialization and its Application in Image Decomposition

Jonathan Brokman, Guy Gilboa

英文摘要：

Branched neural networks have been used extensively for a variety of tasks. Branches are sub-parts of the model that perform independent processing followed by aggregation. It is known that this setting induces a phenomenon called Branch Specialization, where different branches become experts in different sub-tasks. Such observations were qualitative by nature. In this work, we present a methodological analysis of Branch Specialization. We explain the role of gradient descent in this phenomenon. We show that branched generative networks naturally decompose animal images to meaningful channels of fur, whiskers and spots and face images to channels such as different illumination components and face parts.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

阅读原文

阅读 PDF ↗

arXiv 页面 ↗HTML 原文 ↗

本文内容

英文摘要相关论文

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

继续探索

相关论文

使用k稀疏自编码器揭示InceptionV1中的分支专业化

Uncovering Branch specialization in InceptionV1 using k sparse autoencoders

Miramar college（米拉马尔学院）

该研究针对InceptionV1深层分支专业化的未解问题，使用稀疏自编码器揭示了mixed4a-4e、5x5及1x1分支各层的分支专业化示例，并证明其在各层间具有一致性，即相似特征会在对应层相同卷积大小分支中定位。

2025-04-14cs.CV

Diversity-boosted Generalization-Specialization Balancing for Zero-shot Learning

University of New South Wales（新南威尔士大学）；Jiangnan University（江南大学）；University of Technology Sydney（悉尼科技大学）；University of California San Diego（加州大学圣地亚哥分校）；CSIRO’s Data61（澳大利亚联邦科学与工业研究组织Data61）

Zero-Shot Learning (ZSL) aims to transfer classification capability from seen to unseen classes. Recent methods have proved that generalization and specialization are two essential abilities to achieve good performance in ZSL. However, focusing on only one of the abilities may result in models that are either too general with degraded classification ability or too specialized to generalize to unseen classes. In this paper, we propose an end-to-end network, termed as BGSNet, which equips and balances generalization and specialization abilities at the instance and dataset level. Specifically, BG…

2022-10-23cs.CVcs.LG

基于谱分解的ResiDual Transformer对齐

ResiDual Transformer Alignment with Spectral Decomposition

University of Trieste（的里雅斯特大学）；Sapienza University of Rome（罗马第一大学（萨皮恩扎大学））；Institute of Science and Technology Austria (ISTA)（奥地利科学技术研究所）

本文分析视觉Transformer残差流的谱几何特性，发现残差贡献的专业化现象，提出ResiDual技术通过谱对齐放大任务相关成分，在70组实验中实现微调级零样本分类性能。

2025-04-14cs.CVcs.LG

DAF-Net：一种用于红外与可见光图像融合的带域自适应的双分支特征分解融合网络

DAF-Net: A Dual-Branch Feature Decomposition Fusion Network with Domain Adaptive for Infrared and Visible Image Fusion

University of Electronic Science and Technology of China（电子科技大学）；School of Information and Communication Engineering（信息与通信工程学院）；School of Resources and Environment（资源与环境学院）

针对红外与可见光图像融合中关键特征保留难的问题，提出带域自适应的DAF-Net，通过双分支编码器分别提取结构与细节特征，结合MK-MMD对齐特征空间，在多数据集上实现了更优的融合效果。

2024-09-18cs.CVcs.LG

VQ-NeRF：基于矢量量化的神经反射率分解与编辑

VQ-NeRF: Neural Reflectance Decomposition and Editing with Vector Quantization

City University of Hong Kong（香港城市大学）

本文提出VQ-NeRF，通过双分支网络结合矢量量化将三维场景材质离散化，实现低噪声材质分解与交互式离散材质编辑。

2023-11-11cs.CVcs.GRcs.LG

帧级泄露陷阱：重新思考内在图像分解的评估协议，以源分离不确定性为案例研究

The frame-level leakage trap: rethinking evaluation protocols for intrinsic image decomposition, with source-separable uncertainty as a case study

Amazon Web Services (AWS)（亚马逊网络服务（AWS））

本文指出MPI Sintel数据集上学习内在图像分解的评估协议不一致，通过量化帧级泄露效应，提出基于场景级划分的评估标准，并展示源分离不确定性的应用与优势。

2026-05-07cs.CVeess.SP

更多相关论文收起更多论文

Adaptive Frequency Learning in Two-branch Face Forgery Detection

Tsinghua University（清华大学）；Alibaba Security AI Lab（阿里巴巴安全人工智能实验室）

Face forgery has attracted increasing attention in recent applications of computer vision. Existing detection techniques using the two-branch framework benefit a lot from a frequency perspective, yet are restricted by their fixed frequency decomposition and transform. In this paper, we propose to Adaptively learn Frequency information in the two-branch Detection framework, dubbed AFD. To be specific, we automatically learn decomposition in the frequency domain by introducing heterogeneity constraints, and propose an attention-based module to adaptively incorporate frequency features into spati…

2022-03-27cs.CVcs.MM

Semantic decomposition Network with Contrastive and Structural Constraints for Dental Plaque Segmentation

Dalian University of Technology（大连理工大学）

Segmenting dental plaque from images of medical reagent staining provides valuable information for diagnosis and the determination of follow-up treatment plan. However, accurate dental plaque segmentation is a challenging task that requires identifying teeth and dental plaque subjected to semantic-blur regions (i.e., confused boundaries in border regions between teeth and dental plaque) and complex variations of instance shapes, which are not fully addressed by existing methods. Therefore, we propose a semantic decomposition network (SDNet) that introduces two single-task branches to separatel…

2022-08-12cs.AIcs.CV

CDDFuse：用于多模态图像融合的相关性驱动的双分支特征分解

CDDFuse: Correlation-Driven Dual-Branch Feature Decomposition for Multi-Modality Image Fusion

Xi’an Jiaotong University（西安交通大学）；ETH Zürich（苏黎世联邦理工学院）；Northwestern Polytechnical University（西北工业大学）；Research and Development Institute of Northwestern Polytechnical University in Shenzhen（西北工业大学深圳研究院）；Harvard University（哈佛大学）；University of Würzburg（维尔茨堡大学）

提出CDDFuse网络，通过双分支Transformer-CNN提取高低频特征并利用相关性驱动损失进行特征分解，在多模态图像融合及下游任务中取得优异效果。

2023-04-10cs.CV

LaST-SR：用于单图像超分辨率的拉普拉斯启发稳态-瞬态复频分解

LaST-SR: Laplace-Inspired Steady-Transient Complex-Frequency Decomposition for Single Image Super-Resolution

School of Mathematics（数学学院）；Southeast University（东南大学）

本研究针对单图像超分辨率的结构与细节恢复问题，提出LaST-SR模型，通过复频分解与稳态-瞬态协同聚合模块，在×2和×4超分辨率任务的五个基准上实现了最优PSNR/SSIM。

2026-09-03cs.CV