Just wondering how comfortable you guys would feel about the use of Artifical Intelligence within the field of computer forensics? Had an interesting question posed to me today by a student in a uni I work in and not sure how comfortable I would personally be but I think I could see the advantages should it work correctly with thourough testing etc
Oooh me, me !
I think that there is a _lot_ of room for the use of good AI in Forensics, I read AI at Uni for a bit, and I've been dying to put it into practice for a while, time, however has prohibited me thus far. If anyone wants to talk about it though I'm game 😉
Like all things, it has it's benefits and flaws, but in certain areas of triage in large data sets, I think that it might really help to sort the wheat from the chaff …
Il send you a PM azrael
I would feel very comfortable about that ) I've done my masters in AI and really think there are quite a few of areas where AI techniques could be useful - the area that Azrael mentions is a good example but there are other areas where anomaly detection could be useful.
We should be keen to avoid the impression that there will not be any human interaction or decision making involved (and it's easy to imagine some sort of automatic image recognition tool doing all the work - I don't think that's realistic nor desirable).
Just wondering how comfortable you guys would feel about the use of Artifical Intelligence within the field of computer forensics?
Without going into detail
IMHO certain types of AI are very directly applicable, e.g. decision trees, they are transparent in regard to why the AI came to a conclusions. Other types of AI are less transparent, e.g. neural networks, but could still be useful. However these would require additional study about how they came to a conclusion.
I am quite interested in the concept of neural networks. I believe a network that monitors and learns from an investigator can offer real advantages, kind of like a digital triage nurse concept. I am surprised at everyone’s enthusiasm to AI i thought you guys would be very dubious of it.
I am quite interested in the concept of neural networks. I believe a network that monitors and learns from an investigator can offer real advantages, kind of like a digital triage nurse concept. I am surprised at everyone’s enthusiasm to AI i thought you guys would be very dubious of it.
As a CF I remain sceptical, mainly due to the factor acceptance by law, but as a digital researcher I definitely see potential.
The following link is the thesis of Bradley Schatz, which applies AI in combination with semantic web technology in a CF context
http//
If AI is used purely to identify items of interest rather than as an option that produces a report for you then it could be very useful. i.e. use it to draw you attention to a particular file. The onus is then still with the CF guy to determine the relevance.
As an extension to data mining, particularly working over vast data sets, the potential is huge. Relationships could be discovered that an analyst would be unable to find. Or rather so many permutations could be more easily considered, only a positive thing for an investigation.
As azrael stated before, I would definitely take advantage of proven AI for culling of enormous data sets.