Deep-Learning

Artificial Neural Network Practical Exercises

This repository contains several practical exercises on Artificial Neural Networks (ANNs) covering a variety of applications and techniques.

Insurance Cost Prediction

• EDA: Perform exploratory data analysis on the insurance dataset.
• Experimentation: Experiment with Perceptron and different architectures of Deep Neural Networks (DNNs).
• Evaluation: Evaluate model performance using appropriate metrics.
• Hyperparameter Tuning: Experiment with epochs, learning rates, and batch sizes.
• Optimizers: Apply Stochastic Gradient Descent (SGD), Momentum Based GD, and Nesterov Accelerated GD optimizers.

Accident Severity Prediction

• EDA: Perform exploratory data analysis on the accident dataset.
• Evaluation: Evaluate model performance using appropriate metrics.
• Run Time Chart: Display run time chart.
• Hyperparameter Tuning: Experiment with epochs, learning rates, and batch sizes.
• Optimizers: Apply AdaGrad, RMSProp, and Adam optimizers.

Handwritten Digit Recognition

• EDA: Perform exploratory data analysis on the digit classification dataset.
• PCA: Identify appropriate number of principle components and apply PCA.
• Evaluation: Evaluate model performance using appropriate metrics.
• Comparative Analysis: Provide comparative analysis of different models.

Fashion MNIST Image Reconstruction

• Autoencoder Types: Experiment with simple autoencoder and deep autoencoder.
• Activation and Loss Functions: Choose appropriate activation and loss functions for each layer.

Diabetes Prediction

• EDA: Perform exploratory data analysis on the diabetes dataset.
• Regularization: Apply L1, L2, and dropout regularization techniques.
• Evaluation: Evaluate model performance using appropriate metrics.
• Comparative Analysis: Provide comparative analysis of different models.

Dog vs Cat Classification

• Dataset: Download dataset from Kaggle with hierarchy (Train - dog/cat, Test/Val - dog/cat).
• Model: Design and develop a CNN model for classification.

Google Stock Price Prediction using LSTM

• Experimentation: Experiment with split ratio, time stamp value, number of LSTM layers, and neurons in LSTM layer.

Requirements

• Python 3.x
• TensorFlow 2.x
• Keras
• numpy
• matplotlib
• scikit-learn

How to Use

1. Clone the repository.
2. Choose the practical you want to work on.
3. Follow the instructions provided in the respective section of the README.md for that practical.
4. Execute the code and experiment with different parameters and techniques as instructed.