mjc92 / DeepLearningTutorial

Deep Learning tutorial

SK Telecom Deep Learning Lecture Plan

Day 1 (MILAB) 8 hours

Implementing Perceptron

included concepts : Linear models, gradient descent, difference between Batch and Stochastic method, Error backpropagation.

• Rosenblatt Perceptron
• Widrow Hoff Perceptron
• Widrow Hoff + Stochastic Gradient Descent

Implementing Feed-forward neural network

• Simple two layer Neural Network (Numpy implementation)

Day 2 (MILAB + DAVIAN)

Pytorch Basics

Implementing Neural Network with Pytorch

Advanced optimization algorithms (D)

Convolutional Neural Network (D)

Day 3 (DAVIAN)

CNN + Batch norm

Dropout

Data augmentation

popular cnn models

Day 4 (MILAB + DAVIAN)

Transfer learning & Fine-tuning (M)

• Super resolution? or a Network of VGG as feature extractor of MRI + classifier

RNN / LSTM (D)

DAY 5 (DAVIAN)

Attention model (Captioning & Neural machine translation) (D)