Dylan Lema

About

Dylan Lema is a leading researcher at the intersection of federated learning, continual learning, and concept drift adaptation, with a focus on enabling intelligent, decentralized systems on resource-constrained devices. His most influential work, "Concept drift detection and adaptation for federated and continual learning" (2021, 72 citations), tackles a critical challenge in real-world machine learning: how models deployed on smartphones, wearables, and robots can continuously learn from streaming data without forgetting past knowledge or requiring centralized data collection. Lema’s key contribution is the development of robust algorithms that detect and adapt to changing data distributions—concept drift—within federated and continual learning frameworks, ensuring models remain accurate and reliable over time. His research also explores how a "society of devices" can collaboratively perform classification tasks while preserving privacy and minimizing communication overhead. By addressing the practical constraints of mobile and robotic platforms, Lema’s work has significant implications for autonomous systems, smart environments, and edge AI. His publications have laid foundational groundwork for building adaptive, lifelong learning systems that operate effectively in the wild, earning recognition from both the federated learning and robotics communities.

Research Focus

Key Achievements

Top Papers

1. 1
   Concept drift detection and adaptation for federated and continual learning

   72 citations · 2021
2. 2
   Concept Drift Detection and Adaptation for Robotics and Mobile Devices in Federated and Continual Settings

   6 citations · 2020
3. 3
   Federated and continual learning for classification tasks in a society of devices

   6 citations · 2020

Key Collaborators

Contact & Links