Talks

IH&MMSec 2016. I attended the 4th ACM Workshop on Information Hiding & MultiMedia Security, held on June 20-22, 2016, in Vigo, Spain. I was invited to give a keynote talk on "Machine Learning Under Attack: Vulnerability Exploitation and Security Measures", summarizing recent achievements in the area of adversarial machine learning and, in particular, on how sophisticated attacks may target machine learning algorithms and compromise the overall system security. More details and the slides of the presentation can be found here.
​ ICML 2015. I presented our work "Is Feature Selection Secure against Training Data Poisoning?," co-authored with H. Xiao, G. Brown, G. Fumera, C. Eckert and F. Roli, at the International Conference on Machine Learning (ICML 2015, Lille, France, July 6 - 11), one of the top conferences in the area of machine learning.  This is the first work to show that feature selection algorithms can introduce vulnerabilities themselves in security systems, in the presence of attackers that can manipulate even only a small fraction of the training data. The talk is available here.

MCS 2015. I presented the paper "One-and-a-half-class Multiple Classifier Systems for Secure Learning against Evasion Attacks at Test Time," co-authored with I. Corona, Z.-M. He, P. P. K. Chan, G. Giacinto, D. S. Yeung and F. Roli., at the Int'l Workshop on Multiple Classifier Systems (MCS 2015, Guenzburg, Germany, June 29 - July 1). In this work, we showed how to construct a combination of one-class and two-class classifiers to improve system security against attacks aimed at misleading the detection of malicious samples (e.g., spam emails, computer viruses) at test time.

ICB 2015. I presented the paper "Sparse Support Faces," co-authored with Marco Melis, Giorgio Fumera and Fabio Roli, at the International Conference on Biometrics (ICB 2015, Phuket, Thailand, May 19 - 22). Our paper presents a well-principled approach, named sparse support faces, that can outperform state-of-the-art methods both in terms of recognition accuracy and number of required face templates, by jointly learning an optimal combination of matching scores and the corresponding subset of face templates.

MLDM.it 2014. I attended the workshop MLDM.it within the conference AI*IA 2014, held on December 10-11, 2014 in Pisa, Italy. Slides are available here.



 

AISec 2014. I attended the 2014 Workshop on Artificial Intelligence and Security, held on November 7, 2014, in Scottsdale, Arizona, US, within CCS 2014. I gave an invited talk "On Learning and Recognition of Secure Patterns", summarizing our recent achievements in the area of adversarial machine learning, and in particular, on the potential, sophisticated attacks that may target machine learning algorithms to compromise the overall system security. I also presented our recent work on "Poisoning Behavioral Malware Clustering". Some photos of the event are available in our photo gallery.
S+SSPR 2014. I attended the IAPR Joint International Workshop on Statistical, Structural and Syntactic Pattern Recognition, held from Aug. 20-22, 2014 in Joensuu, Finland, where I presented our paper "Poisoning complete-linkage hierarchical clustering". This is a follow-up of our AISec 2013 paper on adversarial clustering.

AISec 2013. I attended the workshop on Artificial Intelligence and Security (AISec 2013), co-located with CCS 2013, one of the major conferences in computer security, held in Berlin, on November, 4 2013, where I presented our work titled  "Is Data Clustering in Adversarial Settings Secure?". The slides are available here. Some pictures of the talk can be found here.  This is our first, seminal work on adversarial clustering.

ECML-PKDD 2013. I attended the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECMLPKDD 2013), held in Prague (CZ) during Sept. 23-27, 2013, where I presented our paper titled "Evasion attacks against machine learning at test time" . The slides are available here.
BTAS 2012. I attended the IEEE Fifth International Conference on Biometrics: Theory, Applications and Systems (BTAS 2012) from Sept. 23 to Sept. 26, 2012, in Washington DC, USA. I presented our recently accepted paper titled "Learning Sparse Kernel Machines with Biometric Similarity Functions for Identity Recognition," co-authored with Giorgio Fumera and Fabio Roli. The talk is available here.

Dagstuhl Workshop on ML for computer security. I attended the Dagstuhl Perspectives Workshop on Machine Learning Methods for Computer Security from Sept. 9 to Sept. 15, 2012, in Dagstuhl, Germany. I gave a 30-min. talk titled "Security evaluation of pattern classifiers: lessons learned in Cagliari". A report of the seminar can be found here.
ICML 2012. I presented the paper "Poisoning attacks against Support Vector Machines," co-authored with Blaine Nelson and Pavel Laskov, at the 29th International Conference on Machine Learning (ICML 2012, Edinburgh, June 25 - July 1). Our paper aims at demonstrating the vulnerability of SVMs (and, more generally, learning algorithms) under adversarial contamination of training data. The talk is available here.