Daily Paper Reading

Daily paper reading records

• ❓ → Plan to read
• ⭐ → Good paper
• ✔️ → Finished for this week
• 📌 → Group meeting talk this week

---

Machine Learning

Surrogate Gap Minimization Improves Sharpness-Aware Training

❓ Open-Set Recognition: A Good Closed-Set Classifier is All You Need

Image as Set of Points

Test-time Adaptation

A Comprehensive Survey on Test-Time Adaptation under Distribution Shifts

A Versatile Framework for Continual Test-Time Domain Adaptation: Balancing Discriminability and Generalizability

Revisiting Test Time Adaptation under Online Evaluation

Evaluating Continual Test-Time Adaptation for Contextual and Semantic Domain Shifts

Test-time Training

ActMAD: Activation Matching to Align Distributions for Test-Time-Training

Test-Time Training with Masked Autoencoders

TTT++: When Does Self-Supervised Test-Time Training Fail or Thrive?

Revisiting Realistic Test-Time Training: Sequential Inference and Adaptation by Anchored Clustering

Test-Time Training with Self-Supervision for Generalization under Distribution Shifts

Test-time Augmentation

❓ Better Aggregation in Test-Time Augmentation

Test-time DG

Test-Time Prompt Tuning for Zero-Shot Generalization in Vision-Language Models

Improved Test-time Adaptation for Domain Generalization

TTA

❓SODA: Robust Training of Test-Time Data Adaptors

ViDA: Homeostatic Visual Domain Adapter for Continual Test Time Adaptation

Towards Open-Set Test-Time Adaptation Utilizing the Wisdom of Crowds in Entropy Minimization

Label Shift Adapter for Test-Time Adaptation under Covariate and Label Shifts

Uncovering Adversarial Risks of Test-Time Adaptation

Gradual Test-Time Adaptation by Self-Training and Style Transfer

❓ Introducing Intermediate Domains for Effective Self-Training during Test-Time

Rethinking Precision of Pseudo Label: Test-Time Adaptation via Complementary Learning Done but haven't been updated

✔️ Adaptive Domain Generalization via Online Disagreement Minimization

✔️ Covariance-aware Feature Alignment with Pre-computed Source Statistics for Test-time Adaptation

✔️ SATA: Source Anchoring and Target Alignment Network for Continual Test Time Adaptation

✔️ CAFA: Class-Aware Feature Alignment for Test-Time Adaptation

❓ Towards Understanding GD with Hard and Conjugate Pseudo-labels for Test-Time Adaptation

✔️ Benchmarking Test-time Unsupervised Deep Neural Network Adaptation on Edge Devices

✔️ Learning to Adapt to Online Streams with Distribution Shifts

✔️ A Simple Test-time Adaptation Method for Source-free Domain Generalization

Robustifying vision transformer without retraining from scratch by test-time class-conditional feature alignment

Test-time batch normalization

On Pitfalls of Test-Time Adaptation

Feature Alignment and Uniformity for Test Time Adaptation

A Probabilistic Framework for Lifelong Test-Time Adaptation

TIPI: Test time adaptation with transformation invariance

EcoTTA: Memory-Efficient Continual Test-time Adaptation via Self-distilled Regularization

Robust mean teacher for continual and gradual test-time adaptation

Multi-step test-time adaptation with entropy minimization and pseudo-labeling

Robust Test-Time Adaptation in Dynamic Scenarios

TeSLA: Test-time self-learning with automatic adversarial augmentation

DELTA: degradation-free fully test-time adaptation

MECTA: Memory-Economic Continual Test-Time Model Adaptation

Parameter-free Online Test-time Adaptation

Decorate the Newcomers: Visual Domain Prompt for Continual Test Time Adaptation

MixNorm: Test-Time Adaptation Through Online Normalization Estimation

Domain Alignment Meets Fully Test-Time Adaptation

Test-time Adaptation via Conjugate Pseudo-Labels

Test-Time Adaptation to Distribution Shifts by Confidence Maximization and Input Transformation

Test-Time Classifier Adjustment Module for Model-Agnostic Domain Generalization

Test-time Batch Statistics Calibration for Covariate Shift

Domain-agnostic Test-time Adaptation by Prototypical Training with Auxiliary Data

Test time Adaptation through Perturbation Robustness

Continual Test-Time Domain Adaptation

Improving Test-Time Adaptation via Shift-agnostic Weight Regularization and Nearest Source Prototypes

MEMO: Test Time Robustness via Adaptation and Augmentation

❓ Extrapolative Continuous-time Bayesian Neural Network for Fast Training-free Test-time Adaptation

TTN: A Domain-Shift Aware Batch Normalization in Test-Time Adaptation

The Norm Must Go On: Dynamic Unsupervised Domain Adaptation by Normalization

Online Adaptation to Label Distribution Shift

Improving robustness against common corruptions by covariate shift adaptation

MM-TTA: Multi-Modal Test-Time Adaptation for 3D Semantic Segmentation

Efficient Test-Time Model Adaptation without Forgetting

❓ Back to the Source: Diffusion-Driven Test-Time Adaptation

Test-Time Adaptation via Self-Training with Nearest Neighbor Information

NOTE: Robust Continual Test-time Adaptation Against Temporal Correlation

Towards Stable Test-time Adaptation in Dynamic Wild World

📌 Neuro-Modulated Hebbian Learning for Fully Test-Time Adaptation

❓ medical Test-Time Unsupervised Domain Adaptation

Imbalanced Data

Learning to Re-weight Examples with Optimal Transport for Imbalanced Classification

❓ ACE: Ally Complementary Experts for Solving Long-Tailed Recognition in One-Shot

✔️ Posterior Re-calibration for Imbalanced Datasets

Transfer Learning things

❓ Does Robustness on ImageNet Transfer to Downstream Tasks?

Domain Adaptation (DA)

Category Contrast for Unsupervised Domain Adaptation in Visual Tasks

Graph-Relational Domain Adaptation

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

Taskonomy: Disentangling Task Transfer Learning

Source-Free Adaptation to Measurement Shift via Bottom-Up Feature Restoration

f-Domain-Adversarial Learning: Theory and Algorithms

⭐ Dirichlet-based Uncertainty Calibration for Active Domain Adaptation

❓ Addressing Parameter Choice Issues in Unsupervised Domain Adaptation by Aggregation

DA for Image Classification

Cycle Self-Training for Domain Adaptation

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

Toalign: Task-oriented alignment for unsupervised domain adaptation

Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks

❓ Upcycling Models under Domain and Category Shift

HyperDomainNet: Universal Domain Adaptation for Generative Adversarial Networks

DA for Semantic Segmentation

Unsupervised Domain Adaptation for Semantic Segmentation using Depth Distribution

Pixel-by-Pixel Cross-Domain Alignment for Few-Shot Semantic Segmentation

TACS: Taxonomy Adaptive Cross-Domain Semantic Segmentation

❓ Domain Transfer through Deep Activation Matching

DA for online 3D segmentation

GIPSO: Geometrically Informed Propagation for Online Adaptation in 3D LiDAR Segmentation

3D point cloud

4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks

3D Semantic Occupancy Prediction

Diffusion

Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Think

Semantic Segmentation

Transformer-based segmentation

Segmenter: Transformer for Semantic Segmentation

Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformers

SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers

Segmentation model

❓ Denoising Pretraining for Semantic Segmentation

❓ Language-driven Semantic Segmentation

Active Boundary Loss for Semantic Segmentation

Segfix: Model-agnostic boundary refinement for segmentation

❓ Segment Anything

Domain Generalization

SWAD: Domain Generalization by Seeking Flat Minima

Domain Generalization by Learning and Removing Domain-specific Features

Ensemble of Averages: Improving Model Selection and Boosting Performance in Domain Generalization

❓ Sparse Mixture-of-Experts are Domain Generalizable Learners

OOD

Assaying Out-Of-Distribution Generalization in Transfer Learning

Visual Prompting via Image Inpainting

Uncertainty Modeling for Out-of-Distribution Generalization

Delving Deep into the Generalization of Vision Transformers under Distribution Shifts

A Fine-Grained Analysis on Distribution Shift

❓ Generalization to Out-of-Distribution transformations

❓ Agree to Disagree: Diversity through Disagreement for Better Transferability

OOD Detection

Mitigating Neural Network Overconfidence with Logit Normalization

❓Beyond AUROC & Co. for Evaluating Out-of-Distribution Detection Performance

❓Decoupling MaxLogit for Out-of-Distribution Detection

PU Learning

Positive-Unlabeled Learning with Non-Negative Risk Estimator

semi-supervised SS

Re-distributing Biased Pseudo Labels for Semi-supervised Semantic Segmentation: A Baseline Investigation

Continual Learning

Learning To Prompt for Continual Learning

Instance Segmentation

SOLO: Segmenting Objects by Locations

SOLOv2: Dynamic and Fast Instance Segmentation

Unsupervised instance segmentation

Freesolo: Learning to segment objects without annotations

Dense Contrastive Learning for Self-Supervised Visual Pre-Training

❓ Cut and Learn for Unsupervised Object Detection and Instance Segmentation

Transformer

❓ Transformers are Sample-Efficient World Models

Calibration

Revisiting the Calibration of Modern Neural Networks

❓ Mitigating Bias in Calibration Error Estimation

Model Selection

❓ LogME: Practical Assessment of Pre-trained Models for Transfer Learning

📌 Transferability Estimation Using Bhattacharyya Class Separability

📌 LEEP: A new measure to evaluate transferability of learned representations

❓ Scalable Diverse Model Selection for Accessible Transfer Learning

❓ Transferability Metrics for Selecting Source Model Ensembles

❓ Ranking and Tuning Pre-trained Models: A New Paradigm for Exploiting Model Hubs

Dataset Distillation

Dataset Distillation by Matching Training Trajectories

Open Vocabulary

A Simple Framework for Open-Vocabulary Segmentation and Detection