BioQ: Towards Context-Aware Multi-Device Collaboration with Bio-cues (ACM SenSys '25)

This is official PyTorch implementation of "BioQ: Towards Context-Aware Multi-Device Collaboration with Bio-cues" (ACM SenSys '25) by Adiba Orzikulova, Diana A. Vasile, Chi Ian Tang, Fahim Kawsar, Sung-Ju Lee, and Chulhong Min.

Tested Environment

• Ubuntu 22.04
• NVIDIA GeForce RTX 3090 GPUs

Step-by-Step Guide

1. Create a new conda environment with required dependencies.

   conda env create -f bioq.yml
   

2. Download and preprocess datasets. To preprocess datasets, refer to the dataset-specific instructions in 00_readme_{dataset_name}.md, located in source/data/data_preprocessor/{dataset_name}.

3. Run experiments. There are two tasks in the paper:

   • Task 1: Cue (embedding) generation
   • Task 2: Cue (embedding) matching

   First, run the cue generation task:

   sh scripts/embgen/{method_name}/embgen_{method_name}_{dataset_name}.sh
   
    # e.g., to run BioQ with wesad dataset:
    sh scripts/embgen/bioq/embgen_bioq_wesad.sh
   

   Then, run the cue matching task:

   sh scripts/matching/{method_name}/matching_{method_name}_{dataset_name}.sh
     
   # e.g., to run BioQ with wesad dataset:
   sh scripts/matching/bioq/matching_bioq_wesad.sh
   

Citation

@inproceedings{orzikulova2025bioq,
  title={BioQ: Towards Context-Aware Multi-Device Collaboration with Bio-cues},
  author={Orzikulova, Adiba and Vasile, Diana A. and Tang, Chi Ian and Kawsar, Fahim and Lee, Sung-Ju and Min, Chulhong},
  booktitle={Proceedings of the 23rd ACM Conference on Embedded Networked Sensor Systems},
  year={2025}
}

---