README

Overview

This repository contains resources for replicating computational experiments with combinatorial CNNs and other architectures focusing on word classification.

Usage Instructions

Setting Up the Environment

1. Datasets:
   • For palindrome-related experiments, download and extract palindromes.zip from this link.
   • Download and extract passwords.zip from the same link for password-related experiments.

Both zip files contain precomputed combinatorics obtained by executing GeneratePalindromes.ipynb, GenerateNonPalindromes.ipynb and GeneratePasswords.ipynb.

Running Experiments

1. Palindrome Experiments:

   • To generate palindromes, use GeneratePalindromes.ipynb.
   • To generate non-palindromes, use GenerateNonPalindromes.ipynb.
   • For (combinatorial) CCNN-based experiments, use palindromeCCNN.ipynb.
   • For CNN-based experiments, use CNN_palindrome.ipynb.
   • For LSTM-based experiments, use LSTM_RNN_palindrome.ipynb.
   • For GRU-based experiments, use GRU_RNN_palindrome.ipynb.
   • For Transformer experiments, use Transformer_palindrome.ipynb.

2. Password Experiments:

   • To generate passwords, use GeneratePasswords.ipynb.
   • For (combinatorial) CCNN-based experiments, use passwordsCCNN.ipynb.
   • For CNN-based experiments, use CNN_passwords.ipynb.
   • For LSTM-based experiments, use LSTM_RNN_passwords.ipynb.
   • For GRU-based experiments, use GRU_RNN_passwords.ipynb.
   • For Transformer experiments, use Transformer_passwords.ipynb.

Additional Resources

• Full lists of palindromes, non-palindromes, strong (good) and weak (bad) passwords are available in corresponding .txt files within the datasets.

License

This project is licensed under the MIT License.