Explainable and transparent AI for Digital Forensics
15 Sep 2021
Harnessing emerging technologies
It is not new to the Digital Forensics field that Artificial Intelligence (AI) techniques are applied to support several activities in the investigation process. However, the techniques being applied so far mainly belong to the area of Machine Learning (ML), i.e., they follow the data-driven and black-box approach in AI which although very efficient suffers from some severe limitations. Currently there is a quest for algorithmic techniques able to guarantee the right to an explanation which is particularly crucial in the legal and forensics domain. Transparency, also, is a desirable feature related to explainability. In this talk I will give a brief overview of ML techniques which are inherently transparent and illustrate some case studies in the forensic domain where they show an added value with respect to traditional ML techniques.