Deep learning homeworks

Repo for homeworks for the deep learning course at IST 2023/24.

Project Structure

The project is structured as follows:

HW1

Homework 1 was about Medical image classification with linear classifiers and neural networks. It was to be implemented using only numpy. There were also some theoretical questions

• HW1/homework1.pdf: This pdf file contains the assignment.

• HW1/hw1-q1.py: Contains code for Question 1, which covers classification with the perception, logistic regression, and the multi-layer perceptron, with implementation in numpy.

• HW1/hw1-q2.py: contains code for Question 2, which covers classification with logistic regression and the multi-layer perceptron, with implementation in torch.

• HW1/report.pdf: This pdf file contains the report for the project, detailing the methods used and the results obtained, as well as answers to the theoretical quetions.

HW2

Homework 2 was about image classification with a CNN and automatic speech recognition, using the attention mechanism. There were also some theoretical questions

• HW2/homework2.pdf: This pdf file contains the assignment of HW 2.

• HW2/data:Subfolder with training data.

• HW2/utils.py: Utility functions for hw2-q2.py

• HW2/hw2-q2.py: Contains code for Question 2, which covers image classification using CNNs, with implementation in torch.

• HW2/hw2-q3.ipnb: Contains code for Question 3, which covers a sequence-to-sequence task using RNNs and attention, with implementation in torch.

• HW2/report.pdf: This pdf file contains the report for the project, detailing the methods used and the results obtained, as well as answers to the theoretical quetions.