This is the codebase for the paper: DNTC: An Unsupervised Deep Networks for Temperature Compensation in Non-stationary Data.
The pooling module incorporates three essential components: Adaptive Average Pooling (AdaptiveAvgPool), Batch Normalization (BatchNorm), and Rectified Linear Unit (ReLU).
The generation module is a Non-stationary Transformer, the input features or sequences undergo multiple transformations of De-stationary Attention.
Its input is provided by both the pooling module and the generation module. This module performs a simple interactive learning operation, and the final reconstructed output provides the compensation signal we need.
- Install Python 3.9 and neccessary dependencies.
pip install -r requirements.txt
- All the 9 datasets can be obtained from Baidu Cloud, Google Drive.
For multivariate compensation results, comparison rate CR of DNTC with baseline methods on the complete dataset.
To evaluate the role of each component in our proposed compensation framework, we removed or replaced the corresponding components and observed how these variations affect the compensation performance on each dataset.
If you have any questions or want to use the code, please contact [email protected].
This repo is built on the Nonstationary_Transformers repo, we appreciate the authors a lot for their valuable code and efforts.