txzhao / Deep-Learning

course assignments of DD2424 deep learning in data science at KTH

Deep-Learning

This is a repo for the individual course assignments of DD2424 Deep Learning in Data Science at KTH 2017. The code in this repo is mainly done in Matlab, and manipulations involved in the training process like gradient calculations and parameter updates are implemented in a trival way (low level).

Datasets

• CIFAR-10 for assignment 1-3

• The Goblet of Fire and Trump Tweets for assignment 4

Contents

• Assignment 1: one-layer network with multiple-class outputs (test accuracy: 40.42%)

  report: basic + optional

• Assignment 2: two-layer network with multiple-class outputs (test accuracy: 54.06%)

  report: basic + optional

• Assignment 3: k-layer network with multiple-class outputs (test accuracy: 54.8%)

  report: basic + optional

• Assignment 4: vanilla RNN to synthesize English text character by character

  report: basic + optional