CSE-343-Machine-Learning

Assignments of the course

Assignment-1

scratch.py

Implemented Linear and Logistic regressions from scratch using only NumPy.

logisticSKlearn.py

Logistic regression implemented using Sklearn.

SGD_Sklearn.py

SGD implementation using Sklearn

EDA.txt

Conducted EDA on Bank Authentication Dataset (in datasets folder)

Assignment-2

Q1.py

Implemented SkLearn's PCA, SVD, Tsne. Did a stratified split and trained a logistic regression using Sklearn.

Q2.py

Used Bootstrapping on the Linear regression model to measure bias and variance of the model.

Q3.py

• Implemented Gridsearch and K-cross validation from scratch and used it on Sklearn's Decision-Trees and GNB classifier.
• Calculated Confusion matrix for binary and multiclass data and plotted their ROCs from scratch

Q4.py

Implemented Gaussian Naive Bayes Classifier from scratch using only NumPy library

Assignment-3

Q1.py

Implemented Multilayer Perceptron with parameters: -

• n layers: Number of Layers (int)
• layer sizes: an array of size n layers which contains the number of nodes in each layer. (array of int)
• activation: activation function to be used (string)
• learning rate: the learning rate to be used (float)
• weight init: initialization function to be used
  • zero: Zero initialization
  • random: Random initialization with a scaling factor of 0.01
  • normal: Normal(0,1) initialization with a scaling factor of 0.01
• batch size: batch size to be used (int)
• num epochs: number of epochs to be used (int)

Q2.py

Used the MLP from Q1 on the MNIST dataset for various hyper-parameter settings

Q3.py

Used PyTorch's MLP for a dataset.

Q4.py

Used pre-trained Alexnet model from PyTorch.