Timezone: »

Balancing Discriminability and Transferability for Source-Free Domain Adaptation
Jogendra Nath Kundu · Akshay Kulkarni · Suvaansh Bhambri · Deepesh Mehta · Shreyas Kulkarni · Varun Jampani · Venkatesh Babu Radhakrishnan

Tue Jul 19 03:30 PM -- 05:30 PM (PDT) @ Hall E #528

Conventional domain adaptation (DA) techniques aim to improve domain transferability by learning domain-invariant representations; while concurrently preserving the task-discriminability knowledge gathered from the labeled source data. However, the requirement of simultaneous access to labeled source and unlabeled target renders them unsuitable for the challenging source-free DA setting. The trivial solution of realizing an effective original to generic domain mapping improves transferability but degrades task discriminability. Upon analyzing the hurdles from both theoretical and empirical standpoints, we derive novel insights to show that a mixup between original and corresponding translated generic samples enhances the discriminability-transferability trade-off while duly respecting the privacy-oriented source-free setting. A simple but effective realization of the proposed insights on top of the existing source-free DA approaches yields state-of-the-art performance with faster convergence. Beyond single-source, we also outperform multi-source prior-arts across both classification and semantic segmentation benchmarks.

Author Information

Jogendra Nath Kundu (Indian Institute of Science)
Akshay Kulkarni (Indian Institute of Science)
Suvaansh Bhambri (Indian Institute of Science)
Deepesh Mehta (Indian Institute of Science)
Shreyas Kulkarni (Indian Institute of Science)
Varun Jampani (Google Research)
Venkatesh Babu Radhakrishnan (Indian Institute of Science)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors