Email Spam Detection is a big issue for most email users and service providers. Email plays a major role in our online communication, like any other medium out there but email is also used to carry spam emails which can be treated as the biggest threat (spam). To build our model we use a dataset of emails with labels (spam or ham) in CSV file format. These methods include batch normalization (BN), dropout, and data augmentation, which let the model generalize better than unsupervised individual transfer learning. While training the model, we update its parameters using gradient descent-based algorithms and evaluate its performance based on accuracy, precision (true positive/true positives + false positives), recall (true positive/ true positives + false negatives), and F1-score. The results confirm that the designed model is reliable and thus very effective in capturing spam emails, providing high precision and recall.

+-