V 0.4.0

• Release structure has features covered for each category/module of neurograph
• Categories are separated based on
  • Data Structures
  • Extraction
  • Graph Library
  • User interface
  • Deep Learning

Data Structures

• Save and load model parameters
  • Integrate these functions with neural network library
• Finish all unit tests for matrice library
• Implement a unique set data structure (tree based approach)

Data Extraction

• Redesign CSV reading and keep track of header and column counts
  • Rewrite frame to graph conversion
• Create module for regular expressions
  • Single match
• Data frame to graph conversion

Deep Learning

• Create Computational graph for back propagation
• Create network with multiple hidden layers

Graph Library

• Community detection
  • Label Propagation (Iterative)
  • Works with up to 2 specific label classes currently
• Add general matrix structure as part of the matrix graph representation
• Triangle count

User Interface

• Expose these functionalities from the graph library
  • Searching algorithms (DFS, BFS)
  • Path Finding (Shortest Path, Dijkstra, Weighted Random Walk)
• Expose these functions from the queue library
  • Initialization
  • Push, Pop and printing queue values
• Demonstrate these in a jupyter notebook

Fixes and Improvements

• Documented steps for all graph theory algorithms
• Cleaned up makefiles
  • Removed repetitive commands in the build process
  • Making use of wildcards for header files
• Ditched the csv library
  • Replaced with generic data frame library