2.2 Classification of the source of samples using spark-induced plasma spectroscopy combined with deep learning - Spark-induced plasma spectroscopy combined with deep learning
- Recently, plasma spectroscopy coupled with supervised machine learning, partial least squares, and artificial neural networks has demonstrated great utilities for efficient classification of samples with similar chemical composition
- This work presents a new attempt on the use of deep learning in identifying the source of fine dust samples
- How to use deep learning in plasma spectroscopy
- The emission spectrum is transformed into black and white images and classified through the feature extraction
- Moreover, the structure and parameters of the two-dimensional CNN model are optimized and determined to be the suitable for the classification
 | | Fig. Deep learning algorithm for classification using SIPS | |
