Research map / Label-efficient and robust learning / Self-supervised and contrastive
Bio-Inspired Self-Supervised Learning for Wrist-worn Accelerometer Data
Prithviraj Tarale, Kiet Chu, Abhishek Varghese, Kai-Chun Liu, Maxwell A Xu, Mohit Iyyer, Sunghoon Lee
This paper tried a common change: Add positional encoding
add positional encodings at transfer time · HAR datasets except WISDM · Macro-F1: 8.2 points on HAD Improved
What 646 accepted papers report for this change
- 525 improved
- 58 no clear effect
- 7 mixed
- 56 hurt
- Learning-Augmented Scalable Linear Assignment Problem Optimization via Neural Dual Warm-Starts ICML 2026 · add positional encodings to features (d=13 -> d=21) · Block-structured, N=4096 · speedupImproved
- Mesh Based Simulations with Spatial and Temporal awareness ICML 2026 · use learned positional embedding, learnable relative bias or weighted adjacency instead of RoPE · Aneurysm · RMSEHurt
- Sparse ActionGen: Accelerating Diffusion Policy with Real-time Pruning ICML 2026 · add sinusoidal positional encoding to the pruner · Robomimic Tool Hang · success rate (%): 28 -> 54Improved
See this change in detailMatched by the wording of each paper's own result, read by AI from the full text. Papers report gains more often than failures, so this shows what gets published, not the chance a change works.
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 changed | Where | Measured | Effect |
|---|---|---|---|
| replace equal-length chunk tokens with movement-segment tokenization (ablation) | Bio-PM · HAR average (6 datasets) | Macro-F1: 0.47 -> 0.65 | Improved |
| add low-frequency gravity residual to the linear probe (ablation) | Bio-PM · HAR average (6 datasets) | Macro-F1: 0.56 -> 0.65 | Improved |
| add SSL pretraining on NHANES (ablation) | Bio-PM · HAR average (6 datasets) | Macro-F1: 0.55 -> 0.65 | Improved |
| movement-segment masked pretraining vs TF-C contrastive | Bio-PM vs TF-C · HAR average (6 datasets) | Macro-F1: 0.59 -> 0.65 | Improved |
| movement-segment pretraining vs augmentation prediction | Bio-PM vs AugPred · HAR average (6 datasets) | Macro-F1: 0.54 -> 0.65 | Improved |
| raise masking rate from 25% to 50% (ablation) | Bio-PM · HAR average (4 datasets) | Macro-F1: 0.65 -> 0.67 | Improved |
| raise masking rate from 50% to 75% (ablation) | Bio-PM · HAR average (4 datasets) | Macro-F1: 0.67 -> 0.65 | Hurt |
| scale model from 1.4M to 6M parameters (ablation) | Bio-PM · HAR average (4 datasets) | Macro-F1: 0.67 -> 0.66 | No clear effect |
| scale model from 1.4M to 13M parameters (ablation) | Bio-PM · HAR average (4 datasets) | Macro-F1: 0.67 -> 0.68 | No clear effect |
| add positional encodings at transfer time (ablation) | Bio-PM · HAR datasets except WISDM | Macro-F1: 8.2 points on HAD | Improved |
| use contextual post-Transformer embeddings vs non-contextual CNN features (ablation) | Bio-PM · unseen-transition next-token prediction | accuracy: +3.3 to +14.9 percentage points | Improved |
| increase additive noise to 3 dB SNR | Bio-PM · HAR with test-time noise | Macro-F1 | Hurt |
Closest papers by meaning
- MoPFormer: Motion-Primitive Transformer for Wearable-Sensor Activity Recognition (NeurIPS 2025 · similarity 0.90)
- Inertia-1: An Open Exploration of Wearable Motion Foundation Models (NeurIPS 2026 · similarity 0.87)
- UniMTS: Unified Pre-training for Motion Time Series (NeurIPS 2024 · similarity 0.87)
- RelCon: Relative Contrastive Learning for a Motion Foundation Model for Wearable Data (ICLR 2025 · similarity 0.87)
- Scaling Wearable Foundation Models (ICLR 2025 · similarity 0.86)
- SensorLM: Learning the Language of Wearable Sensors (NeurIPS 2025 · similarity 0.85)
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.
