From 'Methods of Symmetric Cryptanalysis' by Dmitry Khovratovich,

The data-dependent operations are one of the most controversial design concepts. We say that an operation is data-dependent, if it is expressed as a family of functions indexed by input data. The examples are key-dependent S-boxes, key- and message-dependent rotations and permutations. The advantage of data-dependent operations is that its statistical and diffusion properties are largely unknown to the attacker, so that she can not predict the behaviour of the primitive on a large set of data. On the other hand, the advantage quite often converts to the disadvantage. Indeed, if the statistical property is key-dependent, then by statistical parameters of a sample yield information about the key. The main difficulty of the analysis is hence the extraction of those properties and relating them to a particular key group...

Can we use maching learning techniques for cryptanalysis of these operations?

  • $\begingroup$ Yes There are few papers using genetic algorithms but i would be very curious to know others $\endgroup$ – sashank Dec 4 '14 at 11:16
  • $\begingroup$ @sashank: May I know which are the papers? $\endgroup$ – meta_warrior Dec 4 '14 at 12:03
  • 1
    $\begingroup$ there are plenty , just search for "cryptanalysis genetic algorithms" on scholar.google.com $\endgroup$ – sashank Dec 4 '14 at 15:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.