AI Paper Atlas

Research map / Time series / Foundation models for time series

DeepSenseMoE: Harnessing Power of Time Series Foundation Models for Few-Shot Human Activity Recognition

Zenan Fu, Dongzhou Cheng, Lei Zhang, Wenbo Huang, Zhenghao Chen, Hao Wu

AAAI 2026 · Time series › Foundation models for time series · approach: Time series foundation models · 8 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Add multi-scale convolutional experts with router (case 2 vs case 1) (ablation)DeepSenseMoE · MotionSense 1-shotaccuracy (%): 76.9 -> 80.1Improved
Add shared expert isolation (ablation)DeepSenseMoE · MotionSense 1-shotaccuracy (%): 80.1 -> 81.3Improved
Add hierarchical supervised contrastive alignment (ablation)DeepSenseMoE · MotionSense 1-shotaccuracy (%): 81.3 -> 83.1Improved
Use convolutional experts instead of linear experts (ablation)DeepSenseMoE · MotionSense few-shotaccuracy (%)Improved
Use supervised contrastive instead of Dispersive or InfoNCE (ablation)DeepSenseMoE · MotionSense 5-shotaccuracy (%): 90.2 -> 91.7Improved
Use DeepSenseMoE instead of AdapterMOMENT-small · MotionSense 1-shotaccuracy (%): 68.6 -> 83.1Improved
Use DeepSenseMoE instead of LoRAMOMENT-small · MotionSense 1-shotaccuracy (%): 65.6 -> 83.1Improved
Use DeepSenseMoE instead of full fine-tuningMOMENT-small · PAMAP2 1-shotaccuracy (%): 59.6 -> 74.4Improved
Use DeepSenseMoE instead of Vi2ACT on HHAR 1-shotDeepSenseMoE · HHAR 1-shotaccuracy (%): 75.5 -> 75.1Hurt
Use DeepSenseMoE instead of Vi2ACT on MotionSense 5-shotDeepSenseMoE · MotionSense 5-shotaccuracy (%): 82.2 -> 91.7Improved
Use DeepSenseMoE compared to FT-Head on MotionSense full supervisionDeepSenseMoE · MotionSense Fullaccuracy (%): 98.0 -> 98.9Improved
Vary lambda1 and lambda2 (ablation)DeepSenseMoE · MotionSense 5-shotaccuracy (%)No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy