Table of Contents

• Data Extraction
• Data Structures
• Deep Learning
• Graph Algorithms
• User Interface

Purpose

• Experiment and research algorithms in graph theory and understand it's applications
• Tie concepts from graph theory to deep learning
• Data manipulation tool that utilize graph representations
• Invent a new graph neural network architecture
• Framework to visualize graphs

Design

• Internal map of the core library behind neurograph
• Python library acts as high level user interface for core library
• Allows users to create GNN's and manipulate data
• Build data pipelines with graph theory concepts

flowchart LR

A[Python Library] --> C[CoreLibrary]
C[Core Library] --> B[Graph Theory]
C[Core Library] --> D[Deep Learning]
C[Core Library] --> E[Data Extraction]
C[Core Library] --> F[Data Structures]
D[Deep Learning] --> G[Graph NN's]
B[Graph Theory] --> G[Graph NN's]
E[Data Extraction] --> H[Data Manipulation]
F[Data Structures] --> H[Data Manipulation]

Loading

Dependency Structure

• Dependency structure of the code for neurograph
• Graph and Deep learning library should depend on data structures and extractors
• Data structures and extractors should not depend on the graph or DL library

flowchart LR

A[Graph Library] --> B[Data Structures]
A[Graph Library] --> C[Extractors]
D[Deep Learning] --> B[Data Structures]
D[Deep Learning] --> C[Extractors]

C[Extractors] --> B[Data Structures]

Loading