Intelligent-Systems-Phystech / Shokorov-BS-Thesis

0 stars 0 forks source link

Записки исследования #2

Open vshokorov opened 3 years ago

vshokorov commented 3 years ago

Работы по DDA за последнее время

  1. BDANN : BERT‑based domain adaptation neural network for multi‑modal fake news detection
  2. Generalized Source-free Domain Adaptation
    Click to expand

image

простая идея, схожие объекты должны двигаться рядом при Source-free DA.

  1. Transformer-Based Source-Free Domain Adaptation
    Click to expand

image Техническая работа, прирост за счет архитектуры

  1. Transfer Learning and Deep Domain Adaptation
    • Общий рассказ про то, где находится DA относительно других подходов. Есть некоторая теория, но больше на словах.
  2. Unsupervised Multi-source Domain Adaptation Without Access to Source Data
    Click to expand

image

  1. A Prototype-Oriented Framework for Unsupervised Domain Adaptation
    • Много различных экспериментов, проверки предлагаемого подхода DA.
    • Не самая интересная теория, аккуратно размечают данные статистически ищут prototypes и т.д.
  2. Adversarial Domain Adaptation with Paired Examples for Acoustic Scene Classification on Different Recording Devices
    • Просто применили CycleGAN для source & target датасетов, показали, что качество сильно выросло.
  3. Scaling Laws for the Few-Shot Adaptation of Pre-trained Image Classifiers
    Click to expand

image

  1. Uncertainty-aware Mean Teacher for Source-free Unsupervised Domain Adaptive 3D Object Detection
  2. Unsupervised Cross-Modality Domain Adaptation for Segmenting Vestibular Schwannoma and Cochlea with Data Augmentation and Model Ensemble
  3. Adaptive Sampling using POMDPs with Domain-Specific Considerations
  4. Source-Free Domain Adaptive Fundus Image Segmentation with Denoised Pseudo-Labeling
    Click to expand

image

  1. Unsupervised Domain Adaptation with Semantic Consistency across Heterogeneous Modalities for MRI Prostate Lesion Segmentation
    • Подправили CycleGAN, изменив немного Semantic cycle-consistency loss чтобы иметь возможность преобразовывать целевую переменную (сегментацию).
  2. Powering Comparative Classification with Sentiment Analysis via Domain Adaptive Knowledge Transfer
  3. DILBERT: Customized Pre-Training for Domain Adaptation withCategory Shift, with an Application to Aspect Extraction
  4. Learning Cross-modal Contrastive Features for Video Domain Adaptation
  5. Channel-Temporal Attention for First-Person Video Domain Adaptation
  6. DensePASS: Dense Panoramic Semantic Segmentation via Unsupervised Domain Adaptation with Attention-Augmented Context Exchange
  7. BiMaL: Bijective Maximum Likelihood Approach to Domain Adaptation in Semantic Scene Segmentation
  8. HierMUD: Hierarchical Multi-task Unsupervised Domain Adaptation between Bridges for Drive-by Damage Diagnosis
  9. ⭐️ CLDA: Contrastive Learning for Semi-Supervised Domain Adaptation
  10. Domain adaptation for person re-identification on new unlabeled data using AlignedReID++
  11. Multi-Task Learning and Adapted Knowledge Models for Emotion-Cause Extraction
  12. ⭐️ Predicting the Success of Domain Adaptation in Text Similarity
  13. ⭐️ Correlated Adversarial Joint Discrepancy Adaptation Network
  14. ⭐️ PixMatch: Unsupervised Domain Adaptation via Pixelwise Consistency Training
  15. Incremental Multi-Target Domain Adaptation for Object Detection with Efficient Domain Transfer
vshokorov commented 3 years ago

еще теоретическая работа 2006г. Analysis of Representations for Domain Adaptation