Tom De Schepper

About

Tom De Schepper is a researcher at the forefront of intelligent edge computing and wireless communications, pioneering methods to bridge the gap between advanced machine learning and resource-constrained devices. His work primarily focuses on two transformative areas: spectrum-level traffic classification and efficient deep reinforcement learning (DRL) for low-power systems. In his most cited paper, "Traffic classification at the radio spectrum level using deep learning models trained with synthetic data" (2020, 14 citations), De Schepper introduced a novel approach that bypasses traditional deep packet inspection, enabling traffic recognition directly from radio signals. This method not only enhances privacy but also allows classification without network access—a breakthrough for spectrum monitoring and security. More recently, his 2024 work on "Policy Compression for Intelligent Continuous Control on Low-Power Edge Devices" (3 citations) addresses a critical challenge in deploying DRL on autonomous mobile robots and IoT devices. By compressing complex policies, he enables real-time, reliable inference on hardware with severe power and latency constraints. De Schepper’s contributions are vital for advancing autonomous systems and privacy-preserving network analysis, demonstrating how synthetic data and model compression can unlock new capabilities in edge AI.

Research Focus

Key Achievements

Top Papers

1. 1
   Traffic classification at the radio spectrum level using deep learning models trained with synthetic data

   14 citations · 2020
2. 2
   Policy Compression for Intelligent Continuous Control on Low-Power Edge Devices

   3 citations · 2024

Key Collaborators

Contact & Links