DeepLearningAssignments

Assignments for course Deep Learning, CSE641, IIIT Delhi Winter 22

Assignments

1. Perceptron, Madaline and Multi-Layer Perceptron.

Implement a Perceptron Training Algorithm and plot the decision boundaries for each iteration for AND, OR, NOT and XOR gate.
Implement Adaline and Madaline and train the madaline using the its greedy algorithm for the given data.
Implement a Multi-Layer Perceptron in toolkit.py and optimizers including Gradient Descent, Momentum Optimizer, NAG, RMSProp, AdaGrad and ADAM.

2. CNN and LSTM.

Construct a CNN network to classify cell images into infected or uninfected using Pytorch.
Design and Construct a pipeline using LSTMs to classify dialogues using Pytorch. We also study the impact of increasing context information in classification.

3. Paraphrase Identification using DistilBERT.

Design a pipeline to classify a pair of sentences as paraphrases. We also study the impact of freezing a few layers during training.

Instructions

For each assignment, find the detailed instructions to run each code in the Readme.pdf or Report.pdf. Each question is done in separate Jupyter files (.ipynb).

Details

Course Assignments for Deep Learning (CSE641)

Instructor : Dr. Md. Shad Akhtar

Group Members : Manvi Goel and Eeshaan Ravi Tivari

Skills Used : Machine Learning, Data Cleaning and Visualization, Pytorch, Deep Learning, HuggingFace.