Title: Enhancing sports trainer behaviour monitoring through IoT information processing and advanced deep neural networks

Authors: Zhangying Li; Juan Song

Addresses: Graduate School, Xi'an Physical Education University, Xi'an Shaanxi 710068, China ' Graduate School, Xi'an Physical Education University, Xi'an Shaanxi 710068, China

Abstract: This study investigates the transformative potential of amalgamating IoT technology and deep learning to elevate athletes' behavioural perceptions and revolutionise training programs. The research strategically selects sensors for intelligent wearables, meticulously collects nuanced behavioural data, and employs an innovative semi-supervised ensemble learning approach to handle unlabeled samples. The iterative use of classification information entropy is leveraged to minimise uncertainty, and a deep neural network (DNN) with gradient descent adapts the learning rate, accelerating convergence. The proposed system undergoes rigorous evaluation through cross-validation, demonstrating significant improvements over common methods, including accuracy (93.7%), response time ratio (94.2%), sensitivity ratio (96.2%), and prediction ratio (97.1%). These results underscore the profound impact of the research on advancing athlete training precision, making it a pivotal contribution with broad implications for sports science and performance optimisation.

Keywords: internet of things; IoT; deep neural network; DNN; semi-supervised ensemble learning.

DOI: 10.1504/IJES.2024.143746

International Journal of Embedded Systems, 2024 Vol.17 No.1/2, pp.73 - 84

Received: 22 Dec 2023
Accepted: 21 Mar 2024
Published online: 06 Jan 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article