Deep-Learning

Author: Yutong Wang
Course: Deep Learning (CS324) by Luca Rossi
College: SUSTech

• Perceptron:

  To implement and test a simple artificial neuron: a perceptron.

  • perceptron.ipynb:

• Multiple Layer Perceptron:

  To implement a multi-layer perceptron using Numpy.

  • train_mlp_numpy_BGD.ipynb
  • train_mlp_numpy_SGD.ipynb

  To implement a multi-layer perceptron using Pytorch.

  • mlp_pytorch

• CNN:

  To improve your neural architecture by using convolutional neural networks (CNNs). You will use the CIFAR10 dataset but this time you will implement a CNN to classify the images in the dataset.

  • cnn_train.ipynb

• RNN:

  To implement a RNN to predict the last digit of a palindrome. Palindromes in general are strings that look the same when read forward and backward, e.g., 123454321, 1111, 4224, etc.

  • rnn_train.ipynb

  To implement a LSTM-RNN to predict the last digit of a palindrome.

  • lstm_train.ipynb

• GAN:

  To implement a GAN (see slides for Lesson 9) that generates images similar to those of the training set.

  • my_gan.ipynb