Algorithmic inference using Neural Turing Machines

Final project for University of London, Goldsmith, 2023 by Nikolay Rys.

Main idea

NTM machines to learn regex from data.

Project structure

• The main code is located in the ntm-regex.ipynb notebook that contains the model definitions and experiments.
• The implementation for NTM for Tensorflow 2.0 is located in the ntm.py file.
• datagen.py contains the code for generating data for the experiments.
• In the artefacts folder you can find the saved models.
• In the datasets folder you can find the datasets used for the experiments.

Requirements

• Python 3.9
• Tensorflow 2.4+
• CUDA 11.0+
• cuDNN 8.0+
• automata_toolkit 0.1.0

Licenced as LGPLv3 because the initial code is based on the Neural Turing Machines by Mark Collier