A deep learning model (CNN) build to classify Electron and Photon using matrices provided of the two classes.
The dataset was provided by the CERN. The files name used here are: 'SingleElectronPt50_IMGCROPS_n249k_RHv1.hdf5' and 'SinglePhotonPt50_IMGCROPS_n249k_RHv1.hdf5'.
-
For training set:
- Input: 39,8400 matrices of size 32x32 and 2 channels
- Target: 39,8400 values of 0 or 1
-
For validation set:
- Input: 49800 matrices of size 32x32 and 2 channels
- Target: 49800 values of 0 or 1
-
For test set:
- Input: 49800 matrices of size 32x32 and 2 channels
- Target: 49800 values of 0 or 1
The model and code description are mentioned in the table below:
| DL Framework | Keras |
| Keras version | 2.4.3 |
| Libraries | Numpy, h5py |
| Layers | Conv2D, Batchnormalization, Flatten, Dense |
| Activation functions | ReLU, Sigmoid |
| Optimizer | RMSprop |
| Regularizer | Dropout |
| Learning Rate | 0.0001 |
| Batch size | 32 |
| Epochs | 50 |
| Loss function | Binary Crossentropy |
| Metrics | Accuracy |
| Dataset | Loss | Accuracy |
| Train set | 0.5436 | 0.7344 |
| Validation set | 0.5589 | 0.7235 |
| Test set | 0.5693 | 0.7266 |