東華大學圖書館 |

Deep Learning for Hand Motion Estimation and Biometric Identification Using Surface Electromyography: Enhancing Agility, Generalizability, and Interpretability /

Record Type:	Electronic resources : Monograph/item
Title/Author:	Deep Learning for Hand Motion Estimation and Biometric Identification Using Surface Electromyography: Enhancing Agility, Generalizability, and Interpretability // Qin Hu.
Author:	Hu, Qin,
Description:	1 electronic resource (296 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 86-12, Section: B.
Contained By:	Dissertations Abstracts International86-12B.
Subject:	Electrical engineering. -
Online resource:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=32002145
ISBN:	9798315765356

Deep Learning for Hand Motion Estimation and Biometric Identification Using Surface Electromyography: Enhancing Agility, Generalizability, and Interpretability /
Hu, Qin,

Deep Learning for Hand Motion Estimation and Biometric Identification Using Surface Electromyography: Enhancing Agility, Generalizability, and Interpretability /Qin Hu. - 1 electronic resource (296 pages)

Source: Dissertations Abstracts International, Volume: 86-12, Section: B.

Decoding surface electromyography (sEMG) signals using machine learning methods has attracted substantial research interest over the past two decades due to their non-invasiveness and potential for real-time control applications. sEMG signals-electrical activities produced by muscle contractions-are widely utilized in myoelectric control systems for healthcare applications, including prosthetic and exoskeleton controls, as well as in hands-free interactions within augmented and virtual reality environments.The research presented in this thesis aligns with the broader application of sEMG in human-computer and healthcare interactions, specifically emphasizing model agility, generalizability, and explainable AI (XAI) methodologies. It investigates these themes through hand motion estimation and biometric identification tasks using both traditional sEMG and high-density sEMG (HD-sEMG).Due to inherent neurophysiological differences among individuals and variability among muscle groups within the same individual, sEMG-based hand gesture prediction traditionally requires personalized models. While deep learning significantly outperforms classical machine learning techniques in gesture decoding, the lengthy training processes and substantial data requirements pose practical challenges for real-world implementation. To address these issues, Chapter 2, User-specific Agile Hand Gesture Decoding, introduces optimized, agile deep-learning architectures designed to achieve rapid convergence and robust performance, facilitating efficient decoding of numerous gestures for individual users.Personalized modeling leads to a significant challenge in generalizability, as models trained for one individual perform poorly on others. This lack of transferability severely limits practical usability. To overcome this limitation, Chapter 3, Generalizable Hand Gesture Recognition on Unseen Users, develops strategies for cross-subject gesture prediction, enabling accurate recognition with minimal or even no calibration data from new users.Building upon subject generalizability addressed previously, Chapter 4 tackles the challenge of temporal generalizability. Various practical factors, such as electrode displacement, sensor artifacts, and changes in muscle contraction effort, can significantly degrade model performance over time. To enhance robustness, Reliable Hand Gesture Prediction through Multi-day Evaluation demonstrates methods to maintain high predictive accuracy across multiple days, leveraging few-shot learning.Expanding beyond gesture recognition, sEMG signals also possess unique biometric properties, offering potential advantages for secure human identification and verification. Traditional security methods, including passwords, PINs, and even biometric identifiers such as facial recognition, fingerprints, and iris scans, remain vulnerable to forgery and duplication. In contrast, the neurophysiological nature of sEMG ensures enhanced security due to its inherent uniqueness and the difficulty in duplicating such signals artificially. Chapter 5, sEMG: A New Biometrics Modality for Data Privacy, validates and leverages this novel biometric application to enhance information security.Finally, recognizing the inherent opacity and potential biases associated with deep learning methods, this thesis incorporates XAI methodologies throughout the research. By visualizing and interpreting model predictions through heat maps, XAI techniques illuminate the decision-making processes within deep learning models, ensuring transparency, facilitating model optimization, and ultimately fostering trustworthiness in bio-signal processing algorithms.

English

ISBN: 9798315765356Subjects--Topical Terms:

649834
Electrical engineering.
Subjects--Index Terms:

Biometric identification

Deep Learning for Hand Motion Estimation and Biometric Identification Using Surface Electromyography: Enhancing Agility, Generalizability, and Interpretability /
LDR:05272nmm a22004453i 4500 001 2435049
005 20250930135709.5
006 m o d
007 cr|nu||||||||
008 261011s2025 miu||||||m |||||||eng d
020 $a 9798315765356
035 $a (MiAaPQD)AAI32002145
035 $a AAI32002145
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Hu, Qin, $e author. $3 3825232
245 1 0 $a Deep Learning for Hand Motion Estimation and Biometric Identification Using Surface Electromyography: Enhancing Agility, Generalizability, and Interpretability / $c Qin Hu.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2025
300 $a 1 electronic resource (296 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 86-12, Section: B.
500 $a Advisors: Rangan, Sundeep Committee members: Khorrami, Farsha; Hegde, Chinmay.
502 $b Ph.D. $c New York University Tandon School of Engineering $d 2025.
520 $a Decoding surface electromyography (sEMG) signals using machine learning methods has attracted substantial research interest over the past two decades due to their non-invasiveness and potential for real-time control applications. sEMG signals-electrical activities produced by muscle contractions-are widely utilized in myoelectric control systems for healthcare applications, including prosthetic and exoskeleton controls, as well as in hands-free interactions within augmented and virtual reality environments.The research presented in this thesis aligns with the broader application of sEMG in human-computer and healthcare interactions, specifically emphasizing model agility, generalizability, and explainable AI (XAI) methodologies. It investigates these themes through hand motion estimation and biometric identification tasks using both traditional sEMG and high-density sEMG (HD-sEMG).Due to inherent neurophysiological differences among individuals and variability among muscle groups within the same individual, sEMG-based hand gesture prediction traditionally requires personalized models. While deep learning significantly outperforms classical machine learning techniques in gesture decoding, the lengthy training processes and substantial data requirements pose practical challenges for real-world implementation. To address these issues, Chapter 2, User-specific Agile Hand Gesture Decoding, introduces optimized, agile deep-learning architectures designed to achieve rapid convergence and robust performance, facilitating efficient decoding of numerous gestures for individual users.Personalized modeling leads to a significant challenge in generalizability, as models trained for one individual perform poorly on others. This lack of transferability severely limits practical usability. To overcome this limitation, Chapter 3, Generalizable Hand Gesture Recognition on Unseen Users, develops strategies for cross-subject gesture prediction, enabling accurate recognition with minimal or even no calibration data from new users.Building upon subject generalizability addressed previously, Chapter 4 tackles the challenge of temporal generalizability. Various practical factors, such as electrode displacement, sensor artifacts, and changes in muscle contraction effort, can significantly degrade model performance over time. To enhance robustness, Reliable Hand Gesture Prediction through Multi-day Evaluation demonstrates methods to maintain high predictive accuracy across multiple days, leveraging few-shot learning.Expanding beyond gesture recognition, sEMG signals also possess unique biometric properties, offering potential advantages for secure human identification and verification. Traditional security methods, including passwords, PINs, and even biometric identifiers such as facial recognition, fingerprints, and iris scans, remain vulnerable to forgery and duplication. In contrast, the neurophysiological nature of sEMG ensures enhanced security due to its inherent uniqueness and the difficulty in duplicating such signals artificially. Chapter 5, sEMG: A New Biometrics Modality for Data Privacy, validates and leverages this novel biometric application to enhance information security.Finally, recognizing the inherent opacity and potential biases associated with deep learning methods, this thesis incorporates XAI methodologies throughout the research. By visualizing and interpreting model predictions through heat maps, XAI techniques illuminate the decision-making processes within deep learning models, ensuring transparency, facilitating model optimization, and ultimately fostering trustworthiness in bio-signal processing algorithms.
546 $a English
590 $a School code: 1988
650 4 $a Electrical engineering. $3 649834
650 4 $a Nanoscience. $3 587832
653 $a Biometric identification
653 $a Deep learning
653 $a Hand gesture recognition
653 $a Human computer interaction
653 $a Signal processing
653 $a Surface electromyography
690 $a 0544
690 $a 0800
690 $a 0565
710 2 $a New York University Tandon School of Engineering. $b Electrical & Computer Engineering. $e degree granting institution. $3 3815296
720 1 $a Rangan, Sundeep $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 86-12B.
790 $a 1988
791 $a Ph.D.
792 $a 2025
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=32002145