AI Insight
This paper presents IMU-DM-CLIP, a backdoor attack method that exploits diffusion models to compromise sensor-based human activity recognition (HAR) systems. The researchers demonstrate that by using CLIP-guided diffusion models to generate synthetic training data with embedded triggers, they can successfully attack HAR models that rely on accelerometer and gyroscope data. The attack remains effective even when only 10% of the training data contains backdoor triggers and only 10% of the diffusion model's generation process is guided toward malicious outputs.
Why it matters
This research exposes a critical vulnerability in wearable devices and IoT systems used for health monitoring, fitness tracking, and medical diagnosis. As HAR systems increasingly rely on synthetic data generation to address data scarcity, this backdoor technique reveals how malicious actors could manipulate these systems to produce incorrect activity classifications when specific trigger patterns are present in sensor data.
Understand the Science
arXiv:2606.22837v1 Announce Type: cross
Abstract: Sensors are critical components of modern intelligent devices. The proliferation of the Internet of Things (IoT) and wearable mobile devices has enabled the integration of such sensors to monitor the environment and enable users to take predictive actions. Human activity recognition (HAR) is a popular application in which Inertial Measurement Unit (IMU)-based sensors, such as accelerometers and gyroscopes, are used to provide insights into health, training, and medical diagnosis. However, the accuracy of such a model is hindered by the lack of data. The diffusion model-based technique has proven successful in generating synthetic data for training HAR models. In this paper, we propose a backdoor training technique, IMU-DM-CLIP, that leverages a diffusion model to enable trigger-based attacks on HAR models. Our empirical analysis shows that the attack is successful even with a very small backdoor injection rate of 10% and 10% of the data guided for the diffusion model.
Source: CLIP-guided Diffusion Model for Backdoor Generation in Sensor-based Human Activity Recognition