Hashing has been a crucial technology for digital forensics (DF). It can be used, e.g., to proof integrity of data relied upon as evidence in the court of law, to identify content tampering, to exclude content of no interest, and to detect content of interest. One example application which typically relies on hashing is the detection of Indecent Images of Children (IIOC). It is estimated that, every single day, over 1.8 billion unique images are uploaded and shared online (Microsoft, 2016). A breakthrough in this field was the announcement of a fuzzy hashing algorithm by Microsoft, in partnership with the Dartmouth College, called PhotoDNA (2009). PhotoDNA became the de facto technology used, not only to detect IIOC in DF investigations, but also to remove catalogued IIOC made available online.This talk discusses hashing for detection of IIOC, the role of PhotoDNA, and open challenges in this domain. It also presents results of a study on the classification of PhotoDNA using a dataset of around 3500 images collected from ImageNet and their corresponding PhotoDNA, made available by Microsoft to us. This work became a chapter entitled "Privacy Veriﬁcation of PhotoDNA Based on Machine Learning", part of the book "Security and Privacy for Big Data, Cloud Computing and Applications" published by IET.
Dr Virginia Franqueira is a recently appointed lecturer in cyber security. She is a member of the Kent Interdisciplinary Research Centre in Cyber Security (KirCCS) allocated to the new Institute for Advanced Studies in Cyber Security and Conflict (SoCyETAL). She will be working on planning and design of the MSc Cyber Security (Distance Learning). Previously, she was a senior lecturer at the University of Derby involved in teaching and leadership of programmes related to cyber security and digital forensics (2014-2019), and a lecturer in computing at the University of Central Lancashire (2012-2014). Her research interests in digital forensics span across cybercrime investigation, reconstruction, incident response, and visual content analysis using Machine Learning. In terms of cyber security, her interests lay in cyber security engineering (especially requirements engineering), assessment (e.g., insider threat, IoT security, security of connected & automated vehicles) and management (especially in business networks and computer networks).
Cornwallis South West,
University of Kent,
DetailsOpen to everyone, especially those interested in cyber security research,
Contact: Jason R.C. Nurse