DeepPressNet is a deep learning pipeline that classifies Reddit posts as either "Depressed" or "Not Depressed" based on textual input. It leverages Sentence-BERT (SBERT) for embedding generation and a Multilayer Perceptron (MLP) neural network optimized using Optuna for classification.
Try It Live: Click Here
- End-to-end NLP pipeline
- SBERT for powerful semantic embeddings
- Optuna-based hyperparameter tuning
- Evaluation with precision, recall, F1, ROC AUC, confusion matrix
- CSV logs and .txt reports for model training
- Real-time sentence prediction with confidence score
- Visualizations: training curves, metric comparisons
- SBERT (sentence-transformers)
- TensorFlow / Keras
- Optuna for hyperparameter optimization
- Pandas, NumPy, Scikit-learn, Seaborn, Matplotlib
