Yin ChenNetworked Sensing × Real-world ML

Doctoral research overview

Yin Chen

Networked Sensing × Real-world ML

My research background began in wireless sensor networks and data mining. Over the past decade, I have extended this foundation by designing and operating a residential IoT sensing platform across approximately 95 homes. My doctoral research aims to develop learning methods that remain reliable under cross-home heterogeneity, limited supervision, sensor changes, and long-term distribution shift. I am particularly interested in connecting robust machine learning with the data-generation mechanisms of real-world sensing systems.

Read the problemSee the testbed

Research path

  1. 01

    Identity

    WSN → IoT systems → longitudinal sensing → reliable ML

  2. 02

    Problem

    Reliability under heterogeneity, missing labels, and long-term drift

  3. 03

    Testbed

    ~95 rental homes as a Living Lab / sensing testbed

  4. 04

    Experiments

    Method path sketched; demos & evidence TBD

  5. 05

    About

    Background, continuity, and contact

Asset

A long-term sensing testbed — not a domain cage

A longitudinal residential sensing testbed across ~95 rental homes — a Living Lab for reliable sensing and learning under real deployments.

The irreplaceable asset is the occupied, continuously measured platform — not any single device, and not expertise in one trendy method.

Testbed / Living Lab snapshot

Maps, deployment scale, and data modalities — coming later.

Methods

Tools serve the field — they are not the identity

Anomaly detection, graph signal processing, adaptation, and continual learning are instruments for reliability under real deployments.

Method sketch

Experiment pages will hold interactive demos later (as in prior professor sites).