Symmetries-in-Neural-Networks

This is a repository for my Bachelor's thesis at FEE CTU, called Exploring Symmetries in Deep Learning.

Chapter 1: Functions Invariant to Input Permutations

1.1 Binary Logic Functions

1.1.1 Baseline with Vanilla NNs

1.1.2 NNs with Weight Sharing

Chapter 2: Functions Invariant to Translations

2.1 Convolutional Neural Networks

Chapter 3: Functions Invariant to Input Permutations

3.1 N-Dimensional Logic Functions - 3D XOR

3.2 Set Summation - Deep Sets

Chapter 4: Graph Isomorphisms

3.1 Graph Neural Networks

Chapter 5: Rubik's Cube State Classification

3.1 GNNs for Rubik's Cube State Classification