Publications

Peer-reviewed conference papers, journal articles, and preprints. Author name appears as “Agarwal, H.”

Conference Papers

  1. [1]
    Chakraborty, S., Johri, D., Agarwal, H., Chakraborty, J., Behura, A. “A Hybrid Dual-Head Transformer-Based LSTM Framework for Physiotherapy Exercise Recognition Using Kinect Sensors.” Frontiers in Advanced Computing and Emerging Intelligent Technologies (FACEIT 2026). Learning and Analytics in Intelligent Systems, vol. 63. Springer, Cham. 2026.

    Proposes a hybrid dual-head transformer-based LSTM framework to identify and recognize physiotherapy exercises from spatial-temporal Kinect sensor coordinate inputs.

    [Springer Link]
  2. [2]
    Kumar, Agarwal, H., et al. “Rewiring Transformers for Exploit Likelihood of Vulnerabilities.” AIP Conference Proceedings, ICANTCI 2025. 2025.

    Investigates custom attention mechanics inside transformers (DistilBERT, GPT-2) to predict the likelihood of exploit releases based on National Vulnerability Database (NVD) text reports.

    [AIP Publishing][Semantic Scholar][Wikidata]

Journal Articles (Submitted)

  1. [3]
    Chakraborty, J., Agarwal, H. “Detecting Freezing of Gait (FOG) and Imbalances in Parkinson's Disease Patients.” Health and Technology (Journal). 2025, submitted.

    Utilizes graph spatial neural networks and sequential LSTM modules on wearable sensor streams to predict and identify freezing of gait (FOG) events in patients with Parkinson's disease.