Pith. sign in

REVIEW 18 cited by

Domain Generalization with MixStyle

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2104.02008 v1 pith:MVKERD7W submitted 2021-04-05 cs.CV cs.LG

Kaiyang Zhou , Yongxin Yang , Yu Qiao , Tao Xiang This is my paper · ORCID
classification cs.CVcs.LG
keywords domaindomainsmixstylelearningsourcetrainingdemonstratedgeneralization
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Though convolutional neural networks (CNNs) have demonstrated remarkable ability in learning discriminative features, they often generalize poorly to unseen domains. Domain generalization aims to address this problem by learning from a set of source domains a model that is generalizable to any unseen domain. In this paper, a novel approach is proposed based on probabilistically mixing instance-level feature statistics of training samples across source domains. Our method, termed MixStyle, is motivated by the observation that visual domain is closely related to image style (e.g., photo vs.~sketch images). Such style information is captured by the bottom layers of a CNN where our proposed style-mixing takes place. Mixing styles of training instances results in novel domains being synthesized implicitly, which increase the domain diversity of the source domains, and hence the generalizability of the trained model. MixStyle fits into mini-batch training perfectly and is extremely easy to implement. The effectiveness of MixStyle is demonstrated on a wide range of tasks including category classification, instance retrieval and reinforcement learning.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 18 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. OpenAlex reports about 41 citations worldwide. Full citation record

  1. Quantization Meets OOD: Generalizable Quantization-aware Training from a Flatness Perspective

    cs.CV 2025-08 conditional novelty 7.0 of 10

    Quantization-aware training degrades out-of-distribution accuracy, and a flatness-aware method with gradient-disorder freezing, FQAT, partially recovers it.

  2. Vision Transformer Neural Architecture Search for Out-of-Distribution Generalization: Benchmark and Insights

    cs.LG 2025-01 conditional novelty 7.0 of 10

    A 3000-architecture benchmark shows ViT OoD accuracy varies widely with architecture and that embedding dimension is the strongest structural correlate of OoD robustness.

  3. MoEcho: Exploiting Side-Channel Attacks to Compromise User Privacy in Mixture-of-Experts LLMs

    cs.CR 2025-08 reject novelty 6.0 of 10

    MoEcho claims to compromise user privacy in MoE LLMs and VLMs via four CPU and GPU side channels, but the provided manuscript body contains no supporting content.

  4. CorrMoE: Mixture of Experts with De-stylization Learning for Cross-Scene and Cross-Domain Correspondence Pruning

    cs.CV 2025-07 conditional novelty 6.0 of 10

    CorrMoE combines Progressive Mixstyle de-stylization with a Bi-Fusion Mixture-of-Experts module to improve two-view correspondence pruning in cross-domain and cross-scene settings.

  5. CTA: Cross-Task Alignment for Better Test Time Training

    cs.CV 2025-07 conditional novelty 6.0 of 10

    CTA aligns a SimCLR-trained encoder to a frozen supervised encoder, then adapts only the self-supervised encoder at test time, improving corrupted-image classification over prior test-time training methods.

  6. DAM: Domain-Aware Module for Multi-Domain Dataset Condensation

    cs.CV 2025-05 conditional novelty 6.0 of 10

    A training-time module with learnable spatial masks and frequency-based pseudo-domain labels improves dataset condensation on multi-domain data without increasing images per class.

  7. Seeking Consistent Flat Minima for Better Domain Generalization via Refining Loss Landscapes

    cs.CV 2024-12 conditional novelty 6.0 of 10

    A self-feedback training framework that refines loss landscapes with dynamically generated soft labels finds more consistent flat minima and improves domain generalization accuracy across five benchmarks.

  8. PIER: Physics-Informed Environmental Retrieval for Time-Series Modeling

    cs.LG 2026-07 conditional novelty 5.0 of 10

    PIER augments embedding-based retrieval for lake modeling with a physics-aware stream scored by local verifiers, improving water temperature and dissolved oxygen prediction across 356 lakes.

  9. Towards Realistic Hand-Object Interaction with Gravity-Field Based Diffusion Bridge

    cs.CV 2025-09 conditional novelty 5.0 of 10

    A gravity-field and diffusion-based optimization refines hand-object contacts to reduce interpenetration and gaps while LLM text prompts guide contact regions.

  10. Learning Semantic Directions for Feature Augmentation in Domain-Generalized Medical Segmentation

    cs.CV 2025-07 conditional novelty 5.0 of 10

    A feature-augmentation framework with a learned channel selector and covariance-guided intensity sampler that reports the best average domain-generalized segmentation on Prostate and Fundus benchmarks.

  11. Positive Style Accumulation: A Style Screening and Continuous Utilization Framework for Federated DG-ReID

    cs.CV 2025-07 conditional novelty 5.0 of 10

    SSCU improves federated domain-generalizable person re-identification by screening styles with round-over-round Rank-1 gains and continuously training on the memorized positive styles.

  12. Boosting Domain Generalized and Adaptive Detection with Diffusion Models: Fitness, Generalization, and Transferability

    cs.CV 2025-06 conditional novelty 5.0 of 10

    A single-step diffusion feature extractor with an object-masked auxiliary branch and consistency loss improves domain-generalized and adaptive detection accuracy and speed.

  13. Adaptive Knowledge Distillation using a Device-Aware Teacher for Low-Complexity Acoustic Scene Classification

    cs.SD 2025-09 conditional novelty 4.0 of 10

    A low-complexity scene classifier trained by two-teacher knowledge distillation and device-specific fine-tuning reaches 57.93% accuracy on the DCASE 2025 development set.

  14. Mix, Align, Distil: Reliable Cross-Domain Atypical Mitosis Classification

    cs.CV 2025-08 conditional novelty 4.0 of 10

    A DenseNet-121 trained with MixStyle, CBAM-based feature alignment, and EMA-teacher distillation achieves 0.8762 balanced accuracy on the MIDOG 2025 Task 2 atypical mitosis classification leaderboard.

  15. Fully Automated SAM for Single-source Domain Generalization in Medical Image Segmentation

    cs.CV 2025-07 conditional novelty 4.0 of 10

    FA-SAM automates SAM-based medical segmentation across domains by generating prompt boxes with an uncertainty-enhanced network and fusing image and prompt embeddings.

  16. Fourier Asymmetric Attention on Domain Generalization for Pan-Cancer Drug Response Prediction

    cs.LG 2025-02 reject novelty 4.0 of 10

    FourierDrug uses bulk cell-line expression with adversarial domain generalization and a Fourier asymmetric attention constraint to predict drug response in unseen cancer types, single cells, and patients.

  17. Spatially-Delineated Domain-Adapted AI Classification: An Application for Oncology Data

    cs.LG 2025-01 conditional novelty 4.0 of 10

    A multi-task self-supervised framework with spatial mix-up masking and contrastive predictive coding improves unsupervised domain adaptation for classifying multi-type point maps from cancer tissue regions.

  18. Improving Acoustic Scene Classification in Low-Resource Conditions

    eess.AS 2024-12 conditional novelty 4.0 of 10

    DS-FlexiNet achieves 58.25% accuracy after int8 quantization on TAU22 Task 1A with 30.69K parameters and 8.27M MACs, using residual normalization, ADIR augmentation, and 12-teacher knowledge distillation.

Pith tools