Smaller AI, Smarter Homes: Distilling LLMs for Human Activity Recognition
How can smart homes understand daily activities without heavy, power-hungry AI? A new study tests large language models (LLMs) for Human Activity Recognition (HAR) in homes.
What they found
- Model size matters: Recognition accuracy improves as LLMs get larger.
- Knowledge distillation works: The team fine-tuned smaller LLMs using HAR reasoning examples generated by larger LLMs.
- Big results, small models: Distilled small models performed almost as well as the largest ones while using about 50× fewer parameters.
- Tested on strong benchmarks: Results were shown on two state-of-the-art HAR datasets.
Why it matters: Smaller, high-performing models could make context-aware and assisted-living applications more efficient and accessible.
Paper by Julien Cumin, Oussama Er-Rahmany, and Xi Chen.
Read more: https://arxiv.org/abs/2601.07469v1
Paper: https://arxiv.org/abs/2601.07469v1
Register: https://www.AiFeta.com
AI SmartHome HAR LLM KnowledgeDistillation EdgeAI AssistiveTech MachineLearning Research
