Summary of Efficient Source-free Time-series Adaptation Via Parameter Subspace Disentanglement, by Gaurav Patel et al.
Efficient Source-Free Time-Series Adaptation via Parameter Subspace Disentanglement
by Gaurav Patel, Christopher Sandino, Behrooz Mahasseni, Ellen L Zippi, Erdrin Azemi, Ali Moin, Juri Minxha
First submitted to arxiv on: 3 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Signal Processing (eess.SP)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed framework for Source-Free Domain Adaptation (SFDA) aims to enhance both parameter efficiency and data-sample utilization in time-series adaptation. The approach reparameterizes the source model’s weights using Tucker-style decomposition, factorizing it into a compact form during preparation. A subset of these factors is fine-tuned during target-side adaptation, leading to significant improvements in training efficiency. This selective fine-tuning strategy implicitly regularizes the adaptation process through PAC Bayesian analysis, reducing the overall model size and inference overhead by over 90% while maintaining performance. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper proposes a new way to adapt models from one place to another (domain adaptation) without needing data from that other place. It’s called Source-Free Domain Adaptation (SFDA). The idea is to make it more efficient, using less data and fewer calculations. To do this, the model is broken down into smaller parts, so only some of these parts need to be changed when adapting to a new domain. This makes it faster and uses less resources. The results show that this method can reduce the amount of work needed by over 90% while still getting good results. |
Keywords
* Artificial intelligence * Domain adaptation * Fine tuning * Inference * Time series
