Artificial_Neural_Network_for_Iris_Dataset

This project explores the use of different neural network architectures on the Iris dataset to classify flower species. The dataset is sourced from Kaggle.

Table of Contents:

• Installation
• Usage
• Model Architecture
• Results

Installation:

Ensure you have the required libraries installed:

• numpy
• keras
• scikit-learn
• IPython

You can install them using pip:

  pip install numpy

But mostly in Google colab, these libraries are availabe and requirements are satisfied. But in case, any library does not found, you can downnload it using "pip" statement just shown in above line.

Usage:

Clone this repository to your local machine using this command:

git clone https://github.com/MinahilSadiq1/Artificial_Neural_Network_for_Iris_Dataset.git

At this point

#loading dataset
iris = load_iris()

please provide your dataset file correct name. You can also upload your datafile to google drive and then access it using this piece of code:

iris = '/path_to_your_dataset/yourfilename.csv'

Model-Architecture:

Three neural network architectures are develp in it using different combinations of activation functions: For all three architectures, at Input layer ReLU is used , and for Output layer Softmax is used , but for first architecture, we used Sigmoid activation at hidden layer, for second architecture, we used Tanh at hidden layer. And for third, we used Linear activation function at hidden layer.

1. Sigmoid Model:

   • Input layer with 10 neurons (ReLU activation)
   • Hidden layer with 8 neurons (Sigmoid activation)
   • Output layer with 3 neurons (Softmax activation)

2. Tanh Model:

   • Input layer with 10 neurons (ReLU activation)
   • Hidden layer with 8 neurons (Tanh activation)
   • Output layer with 3 neurons (Softmax activation)

3. Linear Model:

   • Input layer with 10 neurons (ReLU activation)
   • Hidden layer with 8 neurons (Linear activation)
   • Output layer with 3 neurons (Softmax activation)

Results

The models' performance is evaluated on a test set. After training and evaluating each architecture, you should see a summary of the accuracies for each activation function.