# BabyMamba-HAR: Lightweight Selective State Space Models for Efficient Human Activity Recognition on Resource Constrained Devices

arXiv: https://arxiv.org/abs/2602.09872 · HTML: https://mlsanity.com/paper/2602.09872/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Video, motion and generative Mamba models, one of 10 directions in State-space sequence models (S4, Mamba).
- Builds on 14 of its 30 references that are in the map.
- First posted on arXiv 10 Feb 2026.

## Placed in

- State-space sequence models (S4, Mamba) / [Video, motion and generative Mamba models](https://mlsanity.com/ssm/video-motion-and-generative-mamba-models/)

## Builds on (most cited first, 14 of 14)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Mamba: Linear-Time Sequence Modeling with Selective State Spaces](https://mlsanity.com/paper/2312.00752/) (2023)
- [Efficiently Modeling Long Sequences with Structured State Spaces](https://mlsanity.com/paper/2111.00396/) (2021)
- [Deep, Convolutional, and Recurrent Models for Human Activity Recognition using Wearables](https://mlsanity.com/paper/1604.08880/) (2016)
- [Mobile Sensor Data Anonymization](https://mlsanity.com/paper/1810.11546/) (2018)
- [UniMiB SHAR: a new dataset for human activity recognition using acceleration data from smartphones](https://mlsanity.com/paper/1611.07688/) (2016)
- [Deep Learning in Human Activity Recognition with Wearable Sensors: A Review on Advances](https://mlsanity.com/paper/2111.00418/) (2021)
- [On Attention Models for Human Activity Recognition](https://mlsanity.com/paper/1805.07648/) (2018)
- [Improving Deep Learning for HAR with shallow LSTMs](https://mlsanity.com/paper/2108.00702/) (2021)
- [HARMamba: Efficient and Lightweight Wearable Sensor Human Activity Recognition Based on Bidirectional Mamba](https://mlsanity.com/paper/2403.20183/) (2024)
- [TinierHAR: Towards Ultra-Lightweight Deep Learning Models for Efficient Human Activity Recognition on Edge Devices](https://mlsanity.com/paper/2507.07949/) (2025)
- [Beyond Confusion: A Fine-grained Dialectical Examination of Human Activity Recognition Benchmark Datasets](https://mlsanity.com/paper/2412.09037/) (2024)
- [MambaLite-Micro: Memory-Optimized Mamba Inference on MCUs](https://mlsanity.com/paper/2509.05488/) (2025)

## Abstract (arXiv)

Human activity recognition (HAR) on resource constrained devices requires high accuracy across diverse sensor setups. Selective state space models (SSMs) offer efficient linear time sequence processing, presenting a compelling alternative to attention mechanisms. However, their TinyML design space remains unexplored. This paper introduces BabyMamba-HAR, comprising two lightweight architectures: (1) CI-BabyMamba-HAR, utilizing a channel independent stem for noise robustness, and (2) Crossover-BiDir-BabyMamba-HAR, utilizing an early fusion stem for channel count independent complexity. Both integrate weight tied bidirectional scanning and gated temporal attention pooling. Across eight benchmarks, Crossover-BiDir-BabyMamba-HAR averages an 86.52% F1-score with 27K parameters and 2.21M MACs, matching TinyHAR (86.16%) while requiring 11x fewer MACs on high channel datasets. On-device deployment on the Raspberry Pi Pico 2 and ESP32 utilized a mixed precision C++ runtime (INT8 projections, float32 states). A fused computation strategy with lifetime aware memory management reduces peak memory footprint from O(B*dmodel*L*dstate) to O(B*dmodel*dstate), adapting to support weight-tied bidirectional and channel-streaming execution. Both architectures achieved full 8/8 dataset coverage with >99.2% PyTorch parity, whereas INT8 quantized TFLite baselines showed degraded coverage and parity (TinyHAR: 7/8 and 4/8 coverage at 60.4% and 88.6% parity, TinierHAR: 8/8 and 6/8 at 54.2% and 90.8%, DeepConvLSTM: 1/8 and 0/8 on Pico 2 and ESP32, respectively). Crossover-BiDir-BabyMamba-HAR averages 154.4 ms latency on ESP32 and 481.9 ms on Pico 2. Ablations confirm bidirectional scanning and gated attention improve F1-scores by up to 8.42% and 8.94%, respectively, establishing practical principles for TinyML SSM deployment.