PRA Lab works on the development of next generation pattern recognition systems for real applications such as biometric authentication, text categorization, and intrusion detection in computer networks. PRA mission is to address fundamental issues for the development of future pattern recognition systems, in the context of real applications.
Many biometric recognition systems require using a small set of reference templates per client to save computational resources during client verification. PRA Lab's approach, Super-sparse Biometric Recognition, is capable of outperforming state-of-the-art methods both in terms of recognition accuracy and number of required reference templates, by jointly learning an optimal combination of matching scores and the corresponding subset of templates.
PRA Lab has 20 years experience on the development of next-generation Pattern Recognition systems. The Lab Director is Prof. Fabio Roli, IEEE and IAPR fellow. The Lab is made up of more than 30 people, including faculty members, post-doc researchers, PhD students and lab fellows. Research activities are carried out in the framework of regional, national, and european projects funded by public as well as private initiatives. Read more about our researchers.
"There is nothing more practical than a good theory".
Pra Lab develops many tools for computer security. SuStorID is an advanced Intrusion Detection System (IDS) for web services, based on machine learning. It demonstrates a number of interesting features, that can be readily exploited to detect and act against web attacks: Autonomous Learning - Anomaly-based Approach - Multi-model Architecture - Real-time Counteractions - Easy integration with modsecurity - Inspection of Encrypted traffic - User-friendly Interface.
Research at PRA Lab aims to develop secure-by-design systems, natively resilient against the attempts of evasion made by adversaries. The Lab activities focused on the “Adversarial Learning” area aim to study how the learning algorithms that empower our systems can be made more robust against these attempts by proactively simulating an arms race with the adversary to meet more strict security requirements.
Accenture meets PRA Lab
Cagliari, 4 april 2017 – A series of meetings between Accenture and PRA Lab's Biometrics unit has been reported by the Italian press. The second meeting will take place on wednesday 5, April... Complete article (italian only): Unica Notizie and Sardanews.it
The DOGANA literary-graphic competition begins. Contraband Pixels & Texts, or... make stories, not phishing
"Contraband Pixels & Texts, or... make stories, not phishing" is a literary-graphic competition on social engineering and phishing, organized by PRA Lab and CNIT (Consorzio nazionale interuniversitario per le telecomunicazioni), partner of the DOGANA project, funded under the HORIZON 2020 programme.
Participants: writers and cartoonists / illustrators. Registration: registration is free and open to people residing in EU countries (Austria, Belgium, Bulgaria, Cyprus, Croatia, Denmark, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Latvia, Lithuania, Luxembourg , Malta, Netherlands, Poland, Portugal, United Kingdom, Czech Republic, Romania, Slovakia, Slovenia, Spain, Sweden, Hungary), Israel and Switzerland. Rules: writers must submit a short story (max. 5000 characters including spaces, excluding title) addressing the theme provided by the organization;illustrators must submit a table of up to 1024x768 pixels resolution, representing or summarizing the project theme in a drawing or in a comic strip. The same author may submit multiple tables and short stories. Artworks can be submitted in Italian, in English and / or in both languages (Italian and English). Artworks presented in two languages will receive an additional bonus.
Artworks must be shared on the Facebook page dedicated to the competition (https://www.facebook.com/pixelettere), starting from February, 13th, 2017. Last date to submit artworks is June, 10th, 2017. Dates for intermediate selections will be communicated time to time
PRA Lab and Pluribus One among the partners of the challenging European project LETS CROWD, aimed at monitoring and protecting people during mass gatherings. Kick-off meeting: May 11-12, Valencia, Spain.
LETS CROWD (Law Enforcement agencies human factor methods and Toolkit for the Security and protection of CROWDs in mass gatherings) is a project funded by the European Commission under the HORIZON 2020 Programme. Europe has suffered many criminal actions and terrorist attacks during mass gatherings, which have great impact on the citizens and the society, in the last few years. LEAs must face this new scenario (it is considered a priority by the European Union), which imposes a multitude of heterogeneous challenges. Hence, the key is to deter, prevent, protect, pursue and effectively respond to criminal and/or terrorist actions, achieving the best possible protection for people gathering in a specific area where particular events are taking place, thus increasing also the sense of security whit the necessary balance between protection and rights of EU citizens. For all these reasons, novel methodologies and tools must be investigated for strategic and operational activities, involving also strong cross-border cooperation and intelligence sharing, and planning solutions for all these issues, where the human and sociological factor is often the key driver. In fact, humans play a key role in every dimension of crowd protection against criminal and terrorist acts: as perpetrators, protectors and victims.
LETS CROWD will overcome challenges preventing the effective implementation of the European Security Model (ESM) with regards to mass gatherings. This will be achieved by providing the following to security policy practitioners and in particular, LEAs: (1) A dynamic risk assessment methodology for the protection of crowds during mass gatherings centred on human factors in order to effectively produce policies and deploy adequate solutions. (2) A policy making toolkit for the long-term and strategic decision making of security policy makers, including a database of empirical data, statistics and an analytical tool for security policies modelling, and (3) A set of human centred tools for Law Enforcement Agencies (LEAs), including real time crowd behaviour forecasting, innovative communication procedures, semantic intelligence applied to social networks and the internet, and novel computer vision techniques. LETS CROWD will be a security practitioner driven project, fostering the communication and cooperation among LEAs, first responders, civil protection and citizens in the fight against crime and terrorism during mass gatherings by a set of cooperation actions. The project will put citizens at the centre of the research and will assess and evaluate how security measures affect them, and how they perceive them, while respecting EU fundamental rights. LETS CROWD impact will be measured under practical demonstrations involving seven LEAs and relevant emergency services units. In order to facilitate the assessment of the performance, transferability, scalability and large scale deployment of these solutions, the demonstrations will be conducted following eleven use cases.The project, lead by ETRA Investigación y Desarrollo S.A. (Spain), will be implemented by a consortium of 16 partners, from 8 different countries (including SMEs, universities, LEAs), operating in the critical areas of government, security, energy, finance, transport and utilities.
Best thesis CLUSIT 12th edition: Davide Maiorca (phd thesis) and Mauro Deiana (PRA Lab's student) among the winners.
Il CLUSIT published the winners of the "Best thesis nn information security 2016":
1° Tommaso Frassetto (Dipartimento di Matematica, Università degli Studi di Padova) with the thesis: "Self-Rando: Practical Load-Time Randomization against Run-Time Exploits". 2° Leonardo Preti (Politecnico di Milano, Dipartimento di Elettronica, Informazione e Bioingegneria) with the thesis: "Detecting Android malware campaigns via application similarity analysis". 3° Daniele Lain (Dipartimento di Matematica, Università degli Studi di Padova) with the thesis: "Don’t Skype & Type: Keyboard Acoustic Eavesdropping Through Voice Over IP Applications". 4° Marco Negro (Dipartimento di Matematica, Università degli studi di Padova) with the thesis: "Tamper with the flow: Modern Control-Flow Integrity Implementations and their weaknesses". 5° (equal merit) Davide Maiorca (Dipartimento di Ingegneria Elettrica ed Elettronica, Università degli Studi di Cagliari) with the thesis: "Design and Implementation of Robust Systems for Secure Malware Detection". 5° (equal merit) Alessandro Baggio (Politecnico di Milano) with the thesis: "FraudBuster : time-based analysis of Internet banking fraud". 5° (equal merit) Lorenzo Bordoni (Dipartimento di Matematica, Università degli Studi di Padova) with the thesis: "Mirage: Toward a Stealthier and Modular Malware Analysis Sandbox for Android". 5° (equal merit) Mauro Deiana (Dipartimento di Ingegneria Elettrica ed Elettronica, Università degli studi di Cagliari) with the thesis: "Classificazione e rilevazione di malware Android basata sull’analisi del traffico HTTP".
Fabio Roli and Giorgio Giacinto are among the Top italian scientists, according to the ranking published by Via-Academy, specialized site,which made a census and comparison over four thousandresearchers, who have submittedanduploadedtheir works on Googlescholar. From the UNICA's web magazine: Top italian scientists, buone performance dei docenti di UniCa. Il sito specializzato, che ospita i ricercatori che hanno inoltrato e caricato i propri lavori su Google scholar, ha censito oltre quattromila ricercatori. Per l’Università del capoluogo ricadute e visibilità in ambito mondiale. Ventisei i docenti dell’Università di Cagliari nella classifica del sito www.topitalianscientists.org. E potrebbero essercene anche altri: la classifica è in continuo aggiornamento. Di fatto, un bel colpo per l’ateneo del capoluogo sotto diversi punti di vista. Intanto, un risultato proficuo se si pensa alla visibilità, alla credibilità e alle ricadute nella comunità accademica internazionale. Ma anche un bel biglietto da visita in tema di qualità della didattica e della ricerca, utile anche per una promozione puntuale dei corsi di laurea e delle scuole di specializzazione.
ImageCLEF 2017 Lifelog Task, September 11-14, Dublin - Call for participation
This is a cordial invitation to participate in the 1st edition of the Lifelog Task. ImageCLEF is one of the labs of CLEF 2017, which will be held in Dublin, Ireland.
The availability of a large variety of personal devices, such as smartphones, video cameras as well as wearable devices that allow capturing pictures, videos, and audio clips in every moment of our life is creating vast archives of personal data where the totality of an individual's experiences, captured multi-modally through digital sensors are stored permanently as a personal multimedia archive. This unified digital records, commonly referred to as lifelogs, has been gathering increasing attention in recent years within the research community due to the need for systems that can automatically analyse this huge amounts of data in order to categorize, summarize and also query them to retrieve the information that the user may need. Despite the increasing number of successful related workshops and panels in the last years lifelogging has seldom been the subject of a rigorous comparative benchmarking exercise. This task aims to bring the attention of lifelogging to an as wide as possible audience and to promote research into some of the key challenges of the coming years. The task addresses the problems of lifelogging data retrieval and summarization and it is divided in two subtasks based on the same data of a large collection of wearable camera images, description of the semantic locations, and the physical activities of the lifeloggers. The objective of the first subtask is to analyse the lifelog data and, according to several specific queries (e.g., Find the moment(s) when I was shopping for wine in a supermarket), to return the correct answers. The second subtask objective is the analysis of all the images in the dataset and the summarization of them according to specific requirements.
SubTask 1: Lifelog retrieval (LRT) The participants should analyse the lifelog data and according to several specific queries they have to return the correct answers. For example: Shopping for a Bottle of Wine: Find the moment(s) when I was shopping for wine in a supermarket. Shopping For Fish: Find the moment(s) when I was shopping for fish in the supermarket. The Metro: Find the moment(s) when I was riding a metro.
SubTask 2: Lifelog summarization (LST) The participants should analyse all the images and summarize them according to specific requirements. The summary should be represented by 50 images, and it is required to be both relevant and diverse. All of the topics in this subtask will have more than 50 relevant images, so if the participants do not submit 50 images, it will be considered as an incorrect format result. The represented images are considered to be diverse if they depict different moments of the lifelogger in terms of activity, location, day-time, viewpoint, etc. of the queried topic. For example: Public Transport: Summarize the use of public transport by a user. The participant should recognize any different mean of transport depicted in the images of the dataset and if a particular mean of transport it is depicted in different day-time the participant should recognize this.
The participants will have a chance to submit a paper describing their system, which will be published in the CLEF Labs Working Notes. Furthermore, the groups of the best performing systems will be invited to give an oral presentation at CLEF 2017 and others will be given the option of presenting a poster. For more details on the task please visit http://www.imageclef.org/2017/lifelog
Schedule: 14.11.2016: Registration opens. 14.11.2016: Development data release. 20.03.2017: Test data release. 01.05.2017: Deadline for submission of runs by the participants 11:59:59 PM GMT. 15.05.2017: Release of processed results by the task organizers. 26.05.2017: Deadline for submission of working notes papers by the participants 11:59:59 PM GMT. 17.06.2017: Notification of acceptance of the working notes papers. 01.07.2017: Camera ready working notes papers. 11.-14.09.2017: CLEF 2017, Dublin, Ireland.
Organizers: - Duc-Tien Dang-Nguyen, Dublin City University, Ireland (duc-tien.dang-nguyen(at)dcu.ie) - Luca Piras, University of Cagliari, Italy (luca.piras(at)diee.unica.it) - Michael Riegler, University of Oslo, Norway (michaari(at)student.matnat.uio.no) - Cathal Gurrin, Dublin City University, Ireland (cgurrin(at)computing.dcu.ie) - Giulia Boato, University of Trento, Italy (giulia.boato(at)unitn.it)
This work by Luca PirasandGiorgio Giacinto, has been published by the prestigious journalInfomation Fusion, edited byElsevier. From the abstract: An increasing part of communication between persons involve the use of pictures, due to the cheap availability of powerful cameras on smartphones, and storage space. While the real-time use of images is managed through metadata associated with the image (i.e., geolocation), their retrieval from an archive might be far from trivial. After more than 20 years of research on Content-Based Image Retrieval (CBIR), the giant increase in the number and variety of images available in digital format is challenging the research community. Any approach aiming at facing such challenges must rely on different image representations that need to be conveniently fused in order to adapt to the subjectivity of image semantics. This paper offers a journey through the main information fusion ingredients that a recipe for the design of a CBIR system should include to meet the demanding needs of users... Read more >>
ILLBuster: an integrated cyber-intelligence tool for the semi-automatic discovery of illegal activities over the Internet. The system is aimed to be a valuable tool to be used by LEAs for their activities of prevention of and fight against cyber-crime. The developed system consists of: i) a main engine aimed to detect and blacklist malicious internet domains; ii) a set of intelligent peripheral-services that inspect web pages hosted by malicious domains and look for illegal or dangerous material. These intelligent peripheral-services are able to detect child sexual abuse images/videos/material, malware and phishing activities. The resulting system first identifies malicious domains through an analysis of the DNS traffic, then it locates malicious contents hosted by such domains, and finally it reports suspicious URLs to LEAs, so that the illegal domains can be easily identified and persecuted... Read more >>
Some thoughts on safety of Machine Learning. Talk held by Prof. Fabio Roli during HUML 2016 - The Human Use of Machine Learning: An Interdisciplinary Workshop, Venice, December 16th, 2016. From the abstract: "During the last forty years, machine learning research has been focused mainly on accuracy of algorithms, and occasionally on speed and scalability. Issues that are very important in mature engineering fields, like safety, reliability, testing, have been basically neglected. But something changed over the last five years...". Read more >>