Overview / Usage

This project aims at deriving insights into different encryption methods based on their frequency domain. This frequency domain knowledge will then be feed into SVM and Neural Network separately.

In cryptography, an S-box (substitution-box) is a basic component of symmetric key algorithms which performs the substitution. In block ciphers, they are typically used to obscure the relationship between the key and the ciphertext — Shannon's property of confusion.

since I am working with block ciphers all these ciphers have a non-linear unit called an S-Box. so that is why SVM is not working very well because even if projected in higher dimensions these S-Boxes cannot be written as linear functions

Methodology / Approach

I used tfidf of every character in cipher with respect to whole training dataset as our feature vector. In document categorization tasks, a document is considered as a bag of words and it is represented by a document vector whose elements correspond to the frequency of occurrence of different words in the document. The dimension of the document vector is the same as the size of the dictionary built by including all the distinct words that occur in a corpus of documents. Let N be the size of the dictionary. Let ti be the i th word or term in the dictionary, and tf(ti,d) be the frequency of occurrence of ti in a given document d. Then the N-dimensional document vector, φ(d), for a document d is given by: φ(d)=(tf(t1,d), tf(t2,d),..., tf(tN,d))TThis method of representing a text document by a document vector is called as the bag-of-words method. A term or word in a text document corresponds to a sequence of symbols or characters between two consecutive delimiters. we adopted the bag-of-words method for representation of a ciphertext by a document vector.

Technologies Used

intel optimized tensorflow, intel devcloud

Repository

https://github.com/Ujjwal-9/Cipher-Identification