Class-Incremental Domain Adaptation

By Jogendra Nath Kundu, Rahul Venkatesh, Naveen Venkat, Ambareesh Revanur, R. Venkatesh Babu. 2020

58 citations recorded by the source corpus.

We introduce a practical Domain Adaptation (DA) paradigm called Class-Incremental Domain Adaptation (CIDA). Existing DA methods tackle domain-shift but are unsuitable for learning novel target-domain classes. Meanwhile, class-incremental (CI) methods enable learning of new classes in absence of source training data but fail under a domain-shift without labeled supervision. In this work, we effectively identify the limitations of these approaches in the CIDA paradigm. Motivated by theoretical and empirical observations, we propose an effective method, inspired by prototypical networks, that enables classification of target samples into both shared and novel (one-shot) target classes, even under a domain-shift. Our approach yields superior performance as compared to both DA and CI methods in the CIDA paradigm.

Subfields: Artificial Intelligence, Cancer Research, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning, Meta & Continual Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Mingkui Tan
2. Zsolt Kira
3. Stan Sclaroff
4. Kuniaki Saito
5. Chuang Gan
6. Yen-Chang Hsu
7. Yifan Zhang
8. Shuaicheng Niu

In Cancer Research

1. Kuniaki Saito
2. Chuang Gan
3. Kate Saenko
4. Jian Liang
5. Ran He
6. Gabriella Csurka

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. Trevor Darrell
3. Mingkui Tan
4. Andrea Vedaldi
5. Andrew Zisserman
6. Chuang Gan
7. Yi Yang
8. Dacheng Tao

In Representation & Self-Supervised Learning

1. Kuniaki Saito
2. Mingkui Tan
3. Zsolt Kira
4. Stan Sclaroff
5. Yifan Zhang
6. Shuaicheng Niu
7. Chuang Gan
8. Hongbin Lin

In Meta & Continual Learning

1. Zsolt Kira
2. James Smith
3. Yen-Chang Hsu
4. Junjiao Tian
5. Stan Sclaroff
6. Kate Saenko
7. Kai Han
8. Tinne Tuytelaars

Citations in the corpus

• Prototype-Guided Continual Adaptation for Class-Incremental Unsupervised Domain Adaptation (2022, methodology): “Step 3 [20,21,22,23,24,25,26,27,28,29] Postit Notes, Mug, Webcam, Desk Lamp, Telephone, Helmet, Mouse, Pen, Monitor, Mop”
• Recent Advances of Continual Learning in Computer Vision: An Overview (2021, neutral): “As a result, learning the new task can minimally affect the performance of the model on the previous tasks.”
• MgSvF: Multi-Grained Slow versus Fast Framework for Few-Shot Class-Incremental Learning (2020, neutral): “Recently, there has been a large body of research in incremental learning [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43].”
• Universal Domain Adaptation through Self Supervision (2020, methodology): “The result also shows the effectiveness of our method for class-incremental domain adaptation [22].”
• Learning to Classify With Incremental New Class (2021, unknown): “Extra supervision must be introduced to match cluster index to ground-truth labels such as one-shot labeled instances [55].”
• A Closer Look at Rehearsal-Free Continual Learning * (2022, background): “(2)We note here that covariate distribution shifts have been studied in several recent continual learning works (Churamani et al., 2021; Kundu et al., 2020), but this is not the focus of our paper.”
• A Comprehensive Survey on Test-Time Adaptation Under Distribution Shifts (2023, unknown): “To name a few, several recent works [65], [66] consider incremental learning under domain shift.”