3 ms·
This is certainly clever... but it's not about indexing all n-grams of a large size, only the most-frequent k n-grams. Most of my work with n-grams has been fo
by crazygringo 7y ago
This is certainly clever... but it's not about indexing all n-grams of a large size, only the most-frequent k n-grams.
Most of my work with n-grams has been for indexing purposes, meaning you need all the n-grams, not just the most frequent ones.
So if I understand correctly... the application here is to take a large number of programs known to be infected with the same malware, and then run this to find the large chunks in common that will be more reliable as a malware signature in the future.
That's definitely cool. It kind of feels like a different technique from n-grams in the end, though, since presumably just one chunk indicates a true positive? Whereas with n-grams you're generally looking for a fuzzier statistical match?
I'm wondering what other applications this might have besides malware detection.
- dlkf 7y ago> I'm wondering what other applications this might have besides malware detection. I'd be interested to hear if bioinformaticians are interested in this work.
- EdwardRaff 7y agoBioinformatics is one of the application areas we thought this might be useful! I've chatted with a few, and it appears the applications right _now_ might be more limited due to the need for exact matches. It's something we are thinking about for the future, and think it could be useful. There may also be some applications for this in NLP. I'm chatting with a professor who thought this could help speed up some old algorithms that weren't scalable before.
- EdwardRaff 7y agoThanks! Yea, we "cheat" by restricting ourselves to the top-k & distributional assumptions. For our case, the low frequency grams are never used, so it makes sense. >So if I understand correctly... the application here is to take a large number of programs known to be infected with the same malware, and then run this to find the large chunks in common that will be more reliable as a malware signature in the future. You basically got it! We also find large chunks in benign files too. While just _one_ chunk is pretty indicative, performance is much better by using multiple. Thats where Logistic Regression (+ Lasso) comes in to tell us how important each chunk is to making a decision. Near term I think there are some NLP applications for this (though not out to 1024 grams!), and I'm hopeful extensions to this work will be useful to bioinformatics problems. I'm also interested in seeing what craziness people come up with now that large n-grams are an option! Everything I had done/learned before put n>6 as an immediate "why even think about it, you can't compute them" bucket.
- crazygringo 7y agoI've done a lot of work with n-grams in NLP and in my experience it's only useful up to 4-6 words at a time, unless you're trying to index proverbs. The reason being that grammar is intensely hierarchical, so that mere "linear" processing that n-grams do stops being useful beyond things like compound words or short sayings like "don't mind if I do!"
- EdwardRaff 7y agoOh, I totally agree with everything you've said! Some collaborators I'm chatting with have more niche NLP applications where larger n would be valuable though. I don't want to go into too much detail yet since it's their project idea, and not sure on their comfort level on blasting it out to the public yet :)