Disonancias

Main activity

The activity of Tecnológico Fundación Deusto's S³Lab (Laboratory for Smartness, Semantics and Security) concentrates on the application of Artificial Intelligence techniques (expert systems, machine learning, extraction of the essence of knowledge -data mining-, pattern matching, sensitivity analysis, etc.) to the modelling and classification of the behaviour of information systems users. Classification, whose fundamental purpose is to distinguish between normal/legitimate and subversive behaviours.

These techniques have been applied in order to learn automatically (Machine Learning) how users, or the programmes they use, behave in relation to the communications network, with the execution of actions in their operating system (Windows, Linux, etc.) or in their applications in general, and with their interaction with concrete applications (critical applications, in particular), such as economic transactions in e-Banking environments.

So, once a user behaviour profile has been obtained (by means of Data Mining), daily activity is then compared and contrasted with the profile, in a search for significant deviations that might imply a situation of risk for the system. This technique is known as Pattern Matching and it is particularly interesting that it makes it possible to obtain risk estimates, even in systems in which the information is ambiguous or incomplete.

Areas of application and description of the team

S3Lab -Laboratory for Smartness, Semantics and Security- is a laboratory specialising in the application of Artificial Intelligence techniques for modelling problems where it is necessary to deal with the concepts of complexity and certainty.

To date, there have fundamentally been two areas of application:
- Information Security (Cryptanalysis, Network Intrusion Detection, ID at Servers and mobile Terminals, Fraud Detection in Banking, Risk Analysis, Analysis of user and applications behaviour, software virus analysis and detection, etc.);
- Optimisation of Productive Processes (Process modelling, Sensitivity Analysis, Prediction, Simulation, etc.).

The S3Lab is currently formed by 18 researchers, with training in IT Engineering, Telecommunications Engineering, and Mathematics, including four doctors in Computer Science. This combination of profiles is particularly effective when tackling complex problems, generally characterised by the advanced nature of the mathematical support required, located at the frontier of knowledge.

The principal equipment we rely on in the world of Information Security (main area of application) is the grey matter of our researchers, while we do have high performance servers, applicable to problems requiring high computing capacity and main memory. Likewise, we use a KUKA KR6 robotic arm, devoted to the development of conceptual demonstrators of IA algorithms.

Team to be involved with Disonancias

In principle, it is proposed to assign two IT Engineers, with Researcher profiles.

Investigation to be carried out

The fundamental objective pursued is the visualisation of S3Lab’s main line of activity, which is the modelling and classification of user behaviour and/or real time applications, through the above-mentioned mathematical techniques of Artificial Intelligence or others that might be detected as the research activity develops.

A priori, various IA paradigms are currently being used, including Data Mining, which is the extraction of knowledge essence in large data volumes; automatic or Machine Learning; Pattern Matching, which is the ambiguous recognition of patterns -even with incomplete data, or Missing Data-; the adaptation of knowledge to evolutionary changes; behaviour prediction; sensitivity analysis -which is about the interrelation and influence of certain kinds of information over others-, and so forth.

Nevertheless, the technological project might evolve towards other technologies, so its visualisation must be flexible enough to adapt to potential changes in direction. In particular, the concept of Neuronal Network is used as a fundamental mathematical mechanism, although other alternatives could equally be considered at later stages of development, in accordance with the specific ambit of application in each case.

In short, the core of the research proposed is none other than the development of a visualisation tool for the general parallel (distributed) artificial reasoning engine which is being constructed in the S3Lab; a reasoning engine that will have to be capable of modelling anything (extracting essential representative information from a problem) and of inferring classification conclusions in line with its experience within the different areas of application.

Type of exploitation of result

To be determined.

Location

Bilbao (Bizkaia).