Abstract
We demonstrate a machine learning technique for data classification. In particular, we have classified Raman spectral data obtained from extracellular vesicles using a convolutional neural network (CNN).
In this research, 300 spectra from four types of EVs were divided into a training- (60%), validation- (20%) and testing-dataset (20%). Training was performed with the training set and the model is validated with validation set. After the training process, the predictive ability was evaluated with testing set which was not involved in any way during the learning process. We show CNN trained on raw Raman spectra and CNN trained on baseline-corrected Raman data. Classified testing datasets show an accuracy in excess of 90%.
In this research, 300 spectra from four types of EVs were divided into a training- (60%), validation- (20%) and testing-dataset (20%). Training was performed with the training set and the model is validated with validation set. After the training process, the predictive ability was evaluated with testing set which was not involved in any way during the learning process. We show CNN trained on raw Raman spectra and CNN trained on baseline-corrected Raman data. Classified testing datasets show an accuracy in excess of 90%.
| Original language | English |
|---|---|
| Publication status | Published - 30 Aug 2018 |
| Event | XXVI International Conference on Raman Spectroscopy, ICORS 2018 - International Convention Center Jeju, Jeju, Korea, Republic of Duration: 26 Aug 2018 → 31 Aug 2018 Conference number: 26 http://www.icors2018.org |
Conference
| Conference | XXVI International Conference on Raman Spectroscopy, ICORS 2018 |
|---|---|
| Abbreviated title | ICORS |
| Country/Territory | Korea, Republic of |
| City | Jeju |
| Period | 26/08/18 → 31/08/18 |
| Internet address |
Keywords
- Raman Spectroscopy
- Extracellular vesicles
- Label-free biosensing
- Machine Learning
- Convolutional neural networks