GLC++: Source-Free Universal Domain Adaptation Through Global-Local Clustering and Contrastive Affinity Learning (2024-03-21T00:00:00.000000Z)

TL;DR

This paper proposes a novel Global and Local Clustering (GLC) technique, which comprises an adaptive one-vs-all global clustering algorithm to discern between target classes, complemented by a local k-NN clustering strategy to mitigate negative transfer.

Abstract

Deep neural networks often exhibit sub-optimal performance under covariate and category shifts. Source-Free Domain Adaptation (SFDA) presents a promising solution to this dilemma, yet most SFDA approaches are restricted to closed-set scenarios. In this paper, we explore Source-Free Universal Domain Adaptation (SF-UniDA) aiming to accurately classify “known” data belonging to common categories and segregate them from target-private “unknown” data. We propose a novel Global and Local Clustering (GLC) technique, which comprises an adaptive one-vs-all global clustering algorithm to discern between target classes, complemented by a local k-NN clustering strategy to mitigate negative transfer. Despite the effectiveness, the inherent closed-set source architecture leads to uniform treatment of “unknown” data, impeding the identification of distinct “unknown” categories. To address this, we evolve GLC to GLC++, integrating a contrastive affinity learning strategy. We examine the superiority of GLC and GLC++ across multiple benchmarks and category shift scenarios. Remarkably, in the most challenging open-partial-set scenarios, GLC and GLC++ surpass GATE by 16.8% and 18.9% in H-score on VisDA, respectively. GLC++ enhances the novel category clustering accuracy of GLC by 4.1% in open-set scenarios on Office-Home. Furthermore, the introduced contrastive learning strategy not only enhances GLC but also significantly facilitates existing methodologies.

Authors

Florian Rohrbein

3 papers

Changjun Jiang

3 papers

Sanqing Qu

6 papers

TL;DR

Abstract

Authors

References108 items

General Class-Balanced Multicentric Dynamic Prototype Pseudo-Labeling for Source-Free Domain Adaptation

LEAD: Learning Decomposition for Source-free Universal Domain Adaptation

ROAD: Robust Unsupervised Domain Adaptation with Noisy Labels

Crots: Cross-Domain Teacher–Student Learning for Source-Free Domain Adaptive Semantic Segmentation

Upcycling Models Under Domain and Category Shift

PSDC: A Prototype-Based Shared-Dummy Classifier Model for Open-Set Domain Adaptation

Divide and Contrast: Source-free Domain Adaptation via Adaptive Contrastive Learning

Deliberated Domain Bridging for Domain Adaptive Semantic Segmentation

Geometric Anchor Correspondence Mining with Uncertainty Modeling for Universal Domain Adaptation

HRDA: Context-Aware High-Resolution Domain-Adaptive Semantic Segmentation

Reusing the Task-specific Classifier as a Discriminator: Discriminator-free Adversarial Domain Adaptation

BMD: A General Class-balanced Multicentric Dynamic Prototype Strategy for Source-free Domain Adaptation

Towards Robust Adaptive Object Detection under Noisy Annotations

Generalized Category Discovery

UMAD: Universal Model Adaptation under Domain and Category Shift

Generative Adversarial Networks

Source Data-free Unsupervised Domain Adaptation for Semantic Segmentation

Exploiting the Intrinsic Neighborhood Structure for Source-free Domain Adaptation

Model Adaptation: Historical Contrastive Learning for Unsupervised Domain Adaptation without Source Data

Decoupled Adaptation for Cross-Domain Object Detection

Generalize then Adapt: Source-Free Domain Adaptive Semantic Segmentation

Generalized Source-free Domain Adaptation

Novel Visual Category Discovery with Dual Ranking Statistics and Mutual Knowledge Distillation

A Curriculum-Style Self-Training Approach for Source-Free Semantic Segmentation

Source-free Domain Adaptation via Avatar Prototype Generation and Adaptation

Domain Consensus Clustering for Universal Domain Adaptation

Uncertainty Reduction for Model Adaptation in Semantic Segmentation

Iterative human and automated identification of wildlife images

OVANet: One-vs-All Network for Universal Domain Adaptation

Source-Free Domain Adaptation for Semantic Segmentation

Prototypical Pseudo Label Denoising and Target Structure Learning for Domain Adaptive Semantic Segmentation

Source Data-Absent Unsupervised Domain Adaptation Through Hypothesis Transfer and Labeling Transfer

Theoretical Analysis of Self-Training with Deep Networks on Unlabeled Data

On the Effectiveness of Image Rotation for Open Set Domain Adaptation

DACS: Domain Adaptation via Cross-domain Mixed Sampling

Classes Matter: A Fine-grained Adversarial Approach to Cross-domain Semantic Segmentation

Learning From Noisy Labels With Deep Neural Networks: A Survey

Label-Noise Robust Domain Adaptation

Model Adaptation: Unsupervised Domain Adaptation Without Source Data

Source-Relaxed Domain Adaptation for Image Segmentation

Universal Source-Free Domain Adaptation

Towards Inheritable Models for Open-Set Domain Adaptation

A Balanced and Uncertainty-aware Approach for Partial Domain Adaptation

Do We Really Need to Access the Source Data? Source Hypothesis Transfer for Unsupervised Domain Adaptation

Universal Domain Adaptation through Self Supervision

The EU General Data Protection Regulation (GDPR)

A Simple Framework for Contrastive Learning of Visual Representations

Momentum Contrast for Unsupervised Visual Representation Learning

Progressive Domain Adaptation for Object Detection

Learning to Discover Novel Visual Categories via Deep Transfer Clustering

Confidence Regularized Self-Training

Open Set Domain Adaptation: Theoretical Bound and Algorithm

Contrastive Multiview Coding

Universal Domain Adaptation

Separate to Adapt: Open Set Domain Adaptation via Progressive Separation

Transferrable Prototypical Networks for Unsupervised Domain Adaptation

Unsupervised Open Domain Recognition by Semantic Discrepancy Minimization

Bridging Theory and Algorithm for Domain Adaptation

Learning to Transfer Examples for Partial Domain Adaptation

Contrastive Adaptation Network for Unsupervised Domain Adaptation

Multi-class Classification without Multi-class Labels

Moment Matching for Multi-Source Domain Adaptation

ADVENT: Adversarial Entropy Minimization for Domain Adaptation in Semantic Segmentation

Learning deep representations by mutual information estimation and maximization

Partial Adversarial Domain Adaptation

Representation Learning with Contrastive Predictive Coding

Open Set Domain Adaptation by Backpropagation

Maximum Classifier Discrepancy for Unsupervised Domain Adaptation

Image to Image Translation for Domain Adaptation

Learning to cluster in order to Transfer across domains and tasks

CyCADA: Cycle-Consistent Adversarial Domain Adaptation

VisDA: The Visual Domain Adaptation Challenge

Open Set Domain Adaptation

Partial Transfer Learning with Selective Adversarial Networks

Deep Hashing Network for Unsupervised Domain Adaptation

Conditional Adversarial Domain Adaptation