Summary of M3bat: Unsupervised Domain Adaptation For Multimodal Mobile Sensing with Multi-branch Adversarial Training, by Lakmal Meegahapola et al.
M3BAT: Unsupervised Domain Adaptation for Multimodal Mobile Sensing with Multi-Branch Adversarial Training
by Lakmal Meegahapola, Hamza Hassoune, Daniel Gatica-Perez
First submitted to arxiv on: 26 Apr 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
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 A novel approach for multimodal mobile sensing is proposed in this paper, focusing on unsupervised domain adaptation when dealing with multiple modalities of sensor data. The problem of distribution shift, where the training set differs from the real-world deployment environment, is addressed using domain adversarial neural networks (DANN) and a new technique called M3BAT. Extensive experiments are conducted on two datasets, three tasks, and 14 source-target domain pairs, showing that the approach outperforms direct deployment of models trained in the source domain by up to 12% AUC and 0.13 MAE. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper solves a big problem with using mobile sensors to learn about people’s health, behavior, and surroundings. Right now, these systems often don’t work well when they’re used in new places or situations because the data is different from what was trained on. The researchers developed two new techniques that can help adapt to these changes by combining information from multiple types of sensors. They tested their methods on lots of data and showed that they can make models up to 12% more accurate and 0.13% less wrong. |
Keywords
» Artificial intelligence » Auc » Domain adaptation » Mae » Unsupervised
