Deep-Learning-AOT

• Contains homeworks from Deep Learning class of METU CENG

Homeworks are partly adapted from Stanford's cs231n class.

HW1

• Task 1 : SVM exercise (implement SVM loss, lr tuning and regularization, optimize with SVM, visiulizing weights)

  • task 1 notebook

• Task 2 : Softmax exercise (implement softmax loss, lr tuning and regularization, optimize with softmax, visiulizing weights)

  • task 2 notebook

• Task 1 : Implementing a two layer neural network

  • task 3 notebook

• Task 1 : Neural Net for regression (Estimating blood pressure from PPG signal)

  • task 4 notebook

HW2

• Task 1 : Modular neural networks (optimizing the implementation)

  • task 1 notebook

• Task 2 : Training a convolutional network

  • task 2 notebook

• Task 3 : Image gradients and Saliency maps

  • task 3 notebook

• Task 4 a : Neural Net for regression (Estimating blood pressure from PPG signal)

  • task 4 a notebook

• Task 4 b : Image generation from gradients

  • task 4 b notebook

• Task 5 : Estimating blood pressure with CNN architecture

  • task 5 notebook

HW3

• Task 1 : Image Captioning with RNNs

  • task 1 notebook

• Task 2 : Image Captioning with LSTMs

  • task 2 notebook

• Task 3 : Denoising Autoencoder implementation

  • task 3 notebook

• Task 4 : (Optional) Image Captioning with GRUs with GRU basics

  • task 4 notebook