This repository contains the implementation of three types of neural networks for classifying the Iris dataset.
Implemented Neural Networks - Multilayer Perceptron (MLP), Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN). The implementation is done using PyTorch and scikit-learn.
-
Clone the repository:
git clone https://github.com/your-username/Iris-NeuralNetworks-Classification.git
-
Navigate to the project directory:
cd Iris-NeuralNetworks-Classification -
Install the required dependencies:
pip install -r requirements.txt
-
Run the Jupyter Notebook or Python script:
jupyter notebook iris-neural-networks-classification.ipynb
or
python iris-neural-networks-classification.py
The project focuses on implementing and comparing the performance of three neural network architectures—MLP, CNN, and RNN—on the Iris dataset. The Iris dataset is loaded from scikit-learn, and the features are visualized using scatter plots. The data is then preprocessed, split into training and testing sets, and standardized.
-
Multilayer Perceptron (MLP):
- Input Features: 4
- Hidden Layers: 2 (10 neurons each)
- Output Classes: 3
-
Convolutional Neural Network (CNN):
- Input Channels: 1
- Convolutional Layer: 1D with 16 output channels and a filter size of 3
- Max Pooling Layer: 1D with a window size of 2
- Fully Connected Layer: 3 output classes
-
Recurrent Neural Network (RNN):
- LSTM Layer: 1 with 8 hidden units
- Fully Connected Layer: 3 output classes
The project includes training loops for each neural network, visualizing the loss, accuracy, precision, and recall over epochs. Consecutive loss increases are monitored, and early stopping is implemented to prevent overfitting.
To visualize the results, run the provided Jupyter Notebook or Python script.
The implementation includes a sample classification test with new iris samples. The MLP, CNN, and RNN models predict the classes of these samples and print the results.