# CODA: A Continuous Online Evolve Framework for Deploying HAR Sensing Systems

arXiv: https://arxiv.org/abs/2403.14922 · HTML: https://mlsanity.com/paper/2403.14922/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 5 of its 32 references that are in the map.
- First posted on arXiv 22 Mar 2024.

## Builds on (most cited first, 5 of 5)

- [Deep Learning for Sensor-based Human Activity Recognition: Overview, Challenges and Opportunities](https://mlsanity.com/paper/2001.07416/) (2020)
- [Stratified Transfer Learning for Cross-domain Activity Recognition](https://mlsanity.com/paper/1801.00820/) (2017)
- [Semantic-Discriminative Mixup for Generalizable Sensor-based Cross-domain Activity Recognition](https://mlsanity.com/paper/2206.06629/) (2022)
- [Domain Generalization for Activity Recognition via Adaptive Feature Fusion](https://mlsanity.com/paper/2207.11221/) (2022)
- [iMove: Exploring Bio-impedance Sensing for Fitness Activity Recognition](https://mlsanity.com/paper/2402.09445/) (2024)

## Built on by (most cited first, 1 of 1)

- [BayaHAR: Lightweight Bayesian Few-Shot User Adaptation for On-Device Personalized Human Activity Recognition](https://mlsanity.com/paper/2606.04798/) (2026)

## Abstract (arXiv)

In always-on HAR deployments, model accuracy erodes silently as domain shift accumulates over time. Addressing this challenge requires moving beyond one-off updates toward instance-driven adaptation from streaming data. However, continuous adaptation exposes a fundamental tension: systems must selectively learn from informative instances while actively forgetting obsolete ones under long-term, non-stationary drift. To address them, we propose CODA, a continuous online adaptation framework for mobile sensing. CODA introduces two synergistic components: (i) Cache-based Selective Assimilation, which prioritizes informative instances likely to enhance system performance under sparse supervision, and (ii) an Adaptive Temporal Retention Strategy, which enables the system to gradually forget obsolete instances as sensing conditions evolve. By treating adaptation as a principled cache evolution rather than parameter-heavy retraining, CODA maintains high accuracy without model reconfiguration. We conduct extensive evaluations on four heterogeneous datasets spanning phone, watch, and multi-sensor configurations. Results demonstrate that CODA consistently outperforms one-off adaptation under non-stationary drift, remains robust against imperfect feedback, and incurs negligible on-device latency.