DGIST Scholar: Self-Supervised Contrastive Learning Using Temporal-Spectral Hierarchical Loss for Local Field Potentials

Detail View

Department of Electrical Engineering and Computer Science Theses Master

Citations

WEB OF SCIENCE

Citations

SCOPUS

Metadata Downloads

XML

Title: Self-Supervised Contrastive Learning Using Temporal-Spectral Hierarchical Loss for Local Field Potentials

Citation: Sion Kim. (2024). Self-Supervised Contrastive Learning Using Temporal-Spectral Hierarchical Loss for Local Field Potentials. doi: 10.22677/THESIS.200000725263

Description: Self-supervised learning;Contrastive learning;Local field potentials;Hierarchical Contrasting

Table Of Contents: I. Introduction 1
II. Background 5
2.1 Local Field Potentials (LFP) 5
2.2 Self-Supervised Learning 6
2.2.1 Self-Supervised Contrastive Learning 7
2.3 Hierarchical Contrasting 8
III. Research Methods 9
3.1 Problem Definition 9
3.2 Framework Overview 9
3.3 Wavelet Transform 9
3.4 Feature Extraction Encoder 11
3.4.1 Linear Projection Layer 11
3.4.2 Augmentation 12
3.4.3 Stacked Convolutional Blocks 14
3.5 Temporal-Spectral Hierarchical Contrasting 14
IV. Result Analysis 18
4.1 Simulation Datasets 18
4.2 Experimental Concept 18
4.3 Evaluation Metric 19
4.4 Baselines for Comparison 20
4.5 Evaluation Summary 21
4.6 Ablation Study 23
V. Conclusion 25
References 26

URI: http://hdl.handle.net/20.500.11750/48099
http://dgist.dcollection.net/common/orgView/200000725263

Show Full Item Record