Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Applications of Sparse Recovery and Dictionary Learning towards Analysis of Electrodermal Activity

Kelsey, Malia (2017) Applications of Sparse Recovery and Dictionary Learning towards Analysis of Electrodermal Activity. Master's Thesis, University of Pittsburgh. (Unpublished)

Download (1MB) | Preview


Electrodermal Activity (EDA) – a measure of sympathetic nervous system arousal – is one of the primary methods used in psychophysiology and has been used to investigate a variety of physiological topics. EDA signals can be separated into two distinct parts, skin conductance level (SCL) and skin conductance response (SCR). The SCL is a slowly fluctuating response that reflects general trends in activity and level of activation. In contrast SCRs are quick responses which can be seen as peaks in the signal and typically reflect responses to a specific stimulus from the nervous system. While many existing studies collected EDA data in short, laboratory-based experiments, recent developments in wireless biosensing have allowed for out-of-lab studies to become more common. The transition to ambulatory data has introduced challenges in SCR and artifact identification and may hinder analysis of ambulatory data. Therefore, the interest in developing automated systems that can facilitate the analysis of EDA signals has increased in the recent years. Ledalab, a current gold standard, is one such system which can be used to identify SCRs. However, Ledalab is computationally inefficient when applied to long, ambulatory data and does not have the ability to distinguish between SCRs and artifacts. This thesis presents a novel technique that can be used to accurately and efficiently identify SCRs using curve fitting and sparse recovery methods, namely orthogonal matching pursuit. Sparse recovery can also be easily extended to include identification of both SCRs and artifacts given seperability between the two. In this work we apply discriminant analysis to determine if seperability exists between SCRs and artifacts. We have shown that our novel approach was able to detect 69% of the SCRs in an EDA signal compared to the 45% detection of Ledalab. Our method also exponentially decreased the computational complexity compared to the Ledalab. Finally, our system begins addressing the issue of artifact detection by determining that the difference between artifacts and SCR shapes could be distinguished with an accuracy of 74%.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Kelsey, Maliamak341@pitt.eduMAK341
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairAkcakaya,
Committee MemberSejdic,
Committee MemberMao,
Date: 1 February 2017
Date Type: Publication
Defense Date: 28 October 2016
Approval Date: 1 February 2017
Submission Date: 11 November 2016
Access Restriction: 5 year -- Restrict access to University of Pittsburgh for a period of 5 years.
Number of Pages: 101
Institution: University of Pittsburgh
Schools and Programs: Swanson School of Engineering > Electrical Engineering
Degree: MS - Master of Science
Thesis Type: Master's Thesis
Refereed: Yes
Uncontrolled Keywords: Sparse Recovery, Orthogonal Matching Pursuit, Dictionary Learning, Electrodermal Activity, Skin Conductance Response, Artifact Detection
Date Deposited: 01 Feb 2017 17:03
Last Modified: 01 Feb 2022 06:15


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item