Deep Learning Stress Classification Project
Welcome to the Deep Learning Stress Classification project! This repository contains a deep learning model designed to classify a dataset of 369,289 samples into three distinct classes: ‘no stress,’ ‘interruption,’ and ‘time pressure.’ The classification is based on seven input features.
Download Dataset on Kaggle This dataset provides the necessary samples and features for training and testing the model.
The model utilizes Multi-Layer Perceptron (MLP) networks to achieve high accuracy in classification tasks. The architecture consists of the following layers:
- Input Layer: Accepts the seven features.
- First Hidden Layer: 32 neurons.
- Second Hidden Layer: 64 neurons.
- Third Hidden Layer: 32 neurons.
- Output Layer: 3 neurons (one for each class).
- Activation Function: ReLU (Rectified Linear Unit) is employed after each hidden layer to introduce non-linearity into the model.
- Output Layer: The model’s final layer outputs probabilities for each of the three classes.
For training, the model utilizes CrossEntropyLoss from PyTorch, which inherently includes a Softmax layer; hence, a separate Softmax layer is not required in the architecture.
Upon evaluating the model on a test dataset of 41,034 samples, it achieved impressive results:
- Accuracy: Approximately 97%
- Cross Entropy Loss: 0.08
This project demonstrates the effectiveness of MLP networks in classifying stress-related data based on specific features. The high accuracy indicates a strong model performance, making it a valuable tool for applications in psychological research and stress management.
To get started with this project, clone the repository and follow the instructions in the documentation for setting up the environment and running the model.
Thank you for your interest in this project! If you have any questions or suggestions, feel free to reach out.
Email: yassingourkani@outlook.com
LinkedIn: Yasin LinkedIn