Use of Domain Labels during Pre-Training for Domain-Independent WiFi-CSI Gesture Recognition
Bram van Berlo, Richard Verhoeven, Nirvana Meratnia, Yangquan Chen, Nunzio Cennamo, M Jamal Deen, Subhas Mukhopadhyay, Simone Morais, Junseop Lee · 2023
Read Use of Domain Labels during Pre-Training for Domain-Independent WiFi-CSI Gesture Recognition by Bram van Berlo, Richard Verhoeven, Nirvana Meratnia, Yangquan Chen, Nunzio Cennamo, M Jamal Deen, Subhas Mukhopadhyay, Simone Morais, Junseop Lee (2023) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-10675214.txt.
Description
Bram van Berlo and co-authors present a 2023 computing paper on reading hand gestures from WiFi signals. The method senses how gestures disturb the radio waves in a room, so no camera is needed. Models trained in one setting often falter in another, because room traits leak into the signal.
The team adds a domain classifier into the early, unsupervised training phase. This step pushes the model to learn features that hold across settings, and can act as automatic discovery of domain traits. They test single-label and multi-label versions of the classifier on two public data sets, Widar3 and SignFi. Wide cross-checks, leaving whole domains out, gauge how well the trick travels. The work aims at gesture tools that work in new homes and offices without fresh labeling.
Catalog record
| CREATOR | Bram van Berlo, Richard Verhoeven, Nirvana Meratnia, Yangquan Chen, Nunzio Cennamo, M Jamal Deen, Subhas Mukhopadhyay, Simone Morais, Junseop Lee |
|---|---|
| YEAR | 2023 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 103 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |