Continual learning and data analysis of time series data
Time Series Data (TSD) has become the cornerstone of critical applications in various fields. However, temporal analysis faces a significant challenge called catastrophic forgetting, where previously acquired knowledge or skills are lost when learning new tasks. Therefore, this study aims to integra...
محفوظ في:
| المؤلف الرئيسي: | |
|---|---|
| مؤلفون آخرون: | |
| التنسيق: | Final Year Project |
| اللغة: | English |
| منشور في: |
Nanyang Technological University
2024
|
| الموضوعات: | |
| الوصول للمادة أونلاين: | https://hdl.handle.net/10356/176923 |
| الوسوم: |
إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
|
| الملخص: | Time Series Data (TSD) has become the cornerstone of critical applications in various fields. However, temporal analysis faces a significant challenge called catastrophic forgetting, where previously acquired knowledge or skills are lost when learning new tasks. Therefore, this study aims to integrate continuous learning (CL) techniques to mitigate the phenomenon of catastrophic forgetting and enhance the model's capacity for processing TSD. Focusing on the Human Activity Recognition (HAR) problem, this study utilized a hybrid CNN-LSTM hybrid model as the baseline and explored a range of continuous learning techniques, including Experience Replay, Elastic Weight Consolidation (EWC) and Progressive Neural Network (PNN) etc. to show the effectiveness of CL. |
|---|