Abstract
Nondestructive detection methods, based on vibrational spectroscopy, are vitally important in a wide range of applications including industrial chemistry, pharmacy and national defense. Recently, deep learning has been introduced into vibrational spectroscopy showing great potential. Different from images, text, etc. that offer large labeled data sets, vibrational spectroscopic data is very limited, which requires novel concepts beyond transfer and meta learning. To tackle this, we propose a task-enhanced augmentation network (TeaNet). The key component of TeaNet is a reconstruction module that inputs randomly masked spectra and outputs reconstructed samples that are similar to the original ones, but include additional variations learned from the domain. These augmented samples are used to train the classification model. The reconstruction and prediction parts are trained simultaneously, end-to-end with back-propagation. Results on both synthetic and real-world datasets verified the superiority of the proposed method. In the most difficult synthetic scenarios TeaNet outperformed CNN by 17%. We visualized and analysed the neuron responses of TeaNet and CNN, and found that TeaNet's ability to identify discriminant wavenumbers was excellent compared to CNN. Our approach is general and can be easily adapted to other domains, offering a solution to more accurate and interpretable few-shot learning.
Original language | American English |
---|---|
Article number | 10375128 |
Pages (from-to) | 3845-3861 |
Number of pages | 17 |
Journal | IEEE Transactions on Pattern Analysis and Machine Intelligence |
Volume | 46 |
Issue number | 5 |
Early online date | 27 Dec 2023 |
DOIs | |
State | Published - 1 May 2024 |
Keywords
- Masked CNN
- deep learning
- vibrational spectroscopy
All Science Journal Classification (ASJC) codes
- Software
- Artificial Intelligence
- Applied Mathematics
- Computer Vision and Pattern Recognition
- Computational Theory and Mathematics