[论文解读] Density-Aware Graph for Deep Semi-Supervised Visual Recognition

Suichan Li, Bin Liu+4 more|arXiv (Cornell University)|Mar 30, 2020

Domain Adaptation and Few-Shot Learning参考文献 31被引用 6

一句话总结

本文提出了一种密度感知图框架，用于深度半监督视觉识别，通过显式利用数据密度来提升特征学习与伪标签生成。通过引入密度感知邻域聚合（DNA）模块和密度递增路径标签传播（DPLP）模块，该方法实现了端到端训练，在CIFAR-10、CIFAR-100和Mini-ImageNet上均取得了当前最优性能，相较于先前方法显著提升了准确率。

ABSTRACT

Semi-supervised learning (SSL) has been extensively studied to improve the generalization ability of deep neural networks for visual recognition. To involve the unlabelled data, most existing SSL methods are based on common density-based cluster assumption: samples lying in the same high-density region are likely to belong to the same class, including the methods performing consistency regularization or generating pseudo-labels for the unlabelled images. Despite their impressive performance, we argue three limitations exist: 1) Though the density information is demonstrated to be an important clue, they all use it in an implicit way and have not exploited it in depth. 2) For feature learning, they often learn the feature embedding based on the single data sample and ignore the neighborhood information. 3) For label-propagation based pseudo-label generation, it is often done offline and difficult to be end-to-end trained with feature learning. Motivated by these limitations, this paper proposes to solve the SSL problem by building a novel density-aware graph, based on which the neighborhood information can be easily leveraged and the feature learning and label propagation can also be trained in an end-to-end way. Specifically, we first propose a new Density-aware Neighborhood Aggregation(DNA) module to learn more discriminative features by incorporating the neighborhood information in a density-aware manner. Then a novel Density-ascending Path based Label Propagation(DPLP) module is proposed to generate the pseudo-labels for unlabeled samples more efficiently according to the feature distribution characterized by density. Finally, the DNA module and DPLP module evolve and improve each other end-to-end.

研究动机与目标

• 为解决现有半监督学习（SSL）方法中对密度信息利用不足的问题，这些方法将密度信息隐式处理而非显式建模。
• 通过以密度感知方式整合邻域信息，改进特征学习，突破传统基于单一样本的特征聚合方式。
• 通过用在线、基于密度递增路径的方案替代离线、计算密集型方法，实现标签传播的端到端训练。
• 通过一种新颖的密度感知图结构，将特征学习与伪标签生成统一于单一可训练框架中。

提出的方法

• 提出一种密度感知邻域聚合（DNA）模块，基于特征相似性与邻居密度计算注意力权重，在模糊情况下更倾向于高密度邻居。
• 引入一种基于密度递增路径的标签传播（DPLP）模块，通过从已标注或未标注样本出发，沿更高密度邻居逐步构建路径，实现高效的在线标签传播。
• 使用k-NN搜索（例如通过Faiss）构建全局密度感知图，以计算节点密度并为DNA和DPLP定义邻域关系。
• 设计一种联合训练策略，使DNA与DPLP模块端到端协同演化，使特征学习与伪标签生成相互促进。
• 采用密度递增路径约束，基于特征相似性终止路径，确保路径效率与可扩展性。
• 使用Mixup数据增强与标准深度神经网络主干网络，将所提模块整合至标准SSL流程中。

实验结果

研究问题

• RQ1 显式建模数据密度是否能超越隐式密度假设，在半监督视觉识别中带来性能提升？
• RQ2 能否通过密度感知聚合机制有效利用邻域信息进行特征学习？
• RQ3 能否通过构建密度递增路径而非求解大型线性系统，使标签传播更高效且可微分？
• RQ4 与解耦或离线方法相比，端到端训练特征学习与标签传播是否能带来更优性能？

主要发现

• 所提方法在CIFAR-100上使用4,000个标注样本时，达到38.04%的新SOTA错误率，优于先前SOTA 1.94%。
• 在更具挑战性的Mini-ImageNet基准上，该方法在4,000个标注样本下实现55.97%的准确率，超越先前SOTA 14.32%。
• 仅DPLP模块在CIFAR-100（4k标签）上将错误率降低1.54%，仅DNA模块降低1.52%，表明两者均具有显著独立贡献。
• DNA与DPLP联合使用使CIFAR-100（4k标签）错误率降至38.04%，表明两者改进具有互补性。
• 密度递增路径长度分布主要集中在5至25之间，表明在线推理效率高，计算开销可忽略。
• 该框架在推理时间效率上与基线相当，k-NN搜索的开销相比特征提取可忽略不计。