Your WiFi router is constantly monitoring the surrounding. You can analyze the channel state information to detect the location and even trajectory of people in their homes. The majority of these works leverage black box machine learning, which questions their reliability.

While many believe that the black box models provide higher performance and are less complex, new studies suggest otherwise [1]. If you are interested in going against the tide and proving interpretable learning can perform similar to black box models in wireless sensing, send me an email.

Research objective: Explanation methods for WiFi Sensing

Expected gain of knowledge: Wireless communication, Interpretable machine learning

[1] C. Rudin, “Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead,” Nat Mach Intell, vol. 1, no. 5, pp. 206–215, 2019, doi: 10.1038/s42256-019-0048-x.