Contrastive Learning for Time Series Classification with Stronger Augmentations
ID：94
Submission ID：156 View Protection：ATTENDEE
Updated Time：2024-10-23 10:32:53
Hits：50
Oral Presentation
Start Time：2024-11-02 11:10 （Asia/Shanghai）
Duration：20min
Session：[P2] Parallel Session 2 » [P2-2] Parallel Session 2（November 2 AM）
No files
Abstract
How to effectively represent time series data is critically important, as it enables the automatic extraction of features without extensive manual labeling. This capability is crucial in various fields, such as healthcare, where analyzing EEG signals can help diagnose sleep disorders or predict epileptic seizures. In this paper, we propose a novel contrastive learning framework for time series classification, which leverages both strong and weak augmentation strategies to enhance the learning process. We combine time domain and frequency domain augmentation techniques to form a strong augmentation strategy, allowing our model to capture a more comprehensive set of features from time series data. Using a convolutional neural network (CNN) as the backbone for the encoder, our model achieved an accuracy of 83.81% on the Sleep-EDF dataset and 97.73% on the Epilepsy dataset. These results demonstrate the effectiveness of combining strong and weak augmentations for contrastive learning in time series classification tasks.
Keywords
Time series classification,data augmentation,contrastive learning
Speaker
WangZhouyang
student 东南大学Submission Author
All comments
15th August 2024 31st August 2024- Manuscript Submission
15th September 2024 - Acceptance Notification
1st October 2024 - Camera Ready Submission
1st October 2024 – Early Bird Registration
Contact Us
Website:
https://icsmd2024.aconf.org/
Comment submit