Streaming Neural Networks with Memory for Efficient Human Activity Recognition

Research topic/area

Machine Learning, Human-Computer Interaction

Type of thesis

Bachelor

Start time

-

Application deadline

31.03.2027

Duration of the thesis

4 Monate

Description

Human Activity Recognition (HAR) uses motion sensors, such as a smartwatch’s accelerometer, to recognize activities like walking or cooking, often with neural networks running directly on the device. The standard approach cuts the sensor signal into short, overlapping windows and classifies each window on its own. Because the windows overlap, every sensor sample is processed several times, which wastes computation. And because each window is classified in isolation, the network can’t use what happened just before, so predictions often jump between activities. Recent work shows that a memory of previous windows makes predictions more accurate and more consistent [1].

This thesis explores running HAR neural networks on the continuous sensor stream instead, processing each sample only once. This works for purely convolutional networks, which are among the most accurate and efficient HAR models [2, 3, 4]. Their filters compute the same values on the overlapping parts of windows, so one pass over the stream can replace per-window processing [5]. A small recurrent neural network (GRU [6]) on top adds memory across windows. The approach will be applied to existing networks such as CNN-HAR [3] and LiteWay [4], and evaluated on several HAR datasets for both accuracy on complete recordings [1] and computational cost. The goal is a simple, general method that makes HAR neural networks cheaper to run on wearables while giving smoother, more reliable predictions.

References

[1] Bock, M., Gall, J., Moeller, M., & Van Laerhoven, K. “DeepConvContext: A Multi-Scale Approach to Timeseries Classification in Human Activity Recognition.” ISWC, 2026.
[2] HARena: A Living Benchmark for Sensor-Based Human Activity Recognition on Resource-Constrained Edge Devices. Under review.
[3] Yang, J. B., et al. “Deep Convolutional Neural Networks on Multichannel Time Series for Human Activity Recognition.” IJCAI, 2015.
[4] Nshimyimana, D., Fortes Rey, V., Liu, M., Zhou, B., & Lukowicz, P. “LiteWay: Lightweight HAR via Temporal Efficient highWAY.” ISWC, 2026.
[5] Long, J., Shelhamer, E., & Darrell, T. “Fully Convolutional Networks for Semantic Segmentation.” CVPR, 2015.
[6] Cho, K., et al. “Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation.” EMNLP, 2014.

Requirement

Requirements for students

• - Basic knowledge of machine learning and neural networks.
• - Good Python skills; PyTorch experience is a plus.
• - Interest in sensor data, wearables and efficient neural networks.

Faculty departments

• Engineering sciences
  Informatics
  Computer Science

Supervision

Title, first name, last name

Maximilian Burzer, M.Sc.

Organizational unit

TECO, Lehrstuhl für Pervasive Computing Systems (PCS), Institut für Telematik (TM), KIT-Fakultät für Informatik

Email address

maximilian.burzer@kit.edu

Link to personal homepage/personal page

Website

Application via email

Application documents

• Cover letter
• Curriculum vitae
• Grade transcript
• Certificate of enrollment

E-Mail Address for application
Senden Sie die oben genannten Bewerbungsunterlagen bitte per Mail an maximilian.burzer@kit.edu


Back