Lab Assignment-10
Assignment-10- week 4 September
Learning Objective: Learn Neural Networks
Problem Statement
1.Implement a NN from scratch for the XOR classification problem.Use gradient descent. Do not use any library
2.Implement a NN for the XOR problem using tensorflow/keras in python
3.Implement and train a Multilayer Feed-Forward Network (MLP) on the Wine Quality
dataset. Experiment with different numbers of hidden layers and neurons, and discuss how
these choices affect the network’s performance.
Tasks:
● Load and preprocess the Wine Quality dataset.
● Design and implement an MLP with varying architectures (different hidden layers
and neurons).
● Train and evaluate the network.
● Discuss the impact of architecture choices on performance.
4.Implement and compare the performance of a neural network using different activation
functions (Sigmoid, ReLU, Tanh) on the MNIST dataset. Analyze how each activation
function affects the training process and classification accuracy.
Tasks:
● Load and preprocess the MNIST dataset.
● Implement neural networks using Sigmoid, ReLU, and Tanh activation functions.
● Train and evaluate each network.
● Compare training times, convergence, and classification accuracy.
5.Implement and perform hyperparameter tuning for a neural network on the Fashion
MNIST dataset. Experiment with different learning rates, batch sizes, and epochs, and
discuss the impact on model performance.
Tasks:
● Load and preprocess the Fashion MNIST dataset.
● Experiment with different hyperparameters (learning rate, batch size, epochs).
● Train and evaluate the network.
● Discuss how hyperparameter choices affect model performance.
Comments
Post a Comment