CipherSniffer: Classifying Cipher Types

TL;DR

This paper introduces CipherSniffer, a classification approach to identify different cipher types using a dataset of various cipher transformations and evaluating tokenizer-model combinations for effective decryption classification.

Contribution

It presents a novel framing of cipher identification as a classification problem and evaluates multiple models on a comprehensive dataset of cipher types.

Findings

Tokenizer-model combinations vary in classification accuracy.

The dataset includes transpositions, substitutions, reversals, and unencrypted text.

The approach offers a computationally efficient alternative to brute-force decryption.

Abstract

Ciphers are a powerful tool for encrypting communication. There are many different cipher types, which makes it computationally expensive to solve a cipher using brute force. In this paper, we frame the decryption task as a classification problem. We first create a dataset of transpositions, substitutions, text reversals, word reversals, sentence shifts, and unencrypted text. Then, we evaluate the performance of various tokenizer-model combinations on this task.