A fundamental question at the heart of the field of computational collective intelligence is this: how should we aggregate the views of several individual agents so as to arrive at a good representation of the collective view of the group as a whole? In this talk, I will argue that social choice theory, which has classically dealt with the philosophical and mathematical analysis of voting rules used in political elections, can serve as a useful starting point for investigating this question.

The past century, sometimes dubbed the “age of information,” saw incredible growth in computers’ processing speed and memory capacity. Now, since biotechnologies enable massive amounts of data to be collected about living cells and organisms, we have entered into what might be described as an era of the Life Sciences.

A major challenge in the analysis of such data sets is their size: a very small number of observations (records), of the order of tens or hundreds, versus thousands (or more!) of attributes or features per each observation (such problems are now dubbed small-n-large-p-problems, where n stands for the sample size and p denotes the observation vector’s dimension). Typical examples include microarray gene expression experiments (where the features are gene expression levels) or data coming from genome-wide association studies (where the features are the so-called genetic markers, e.g., single nucleotide polymorphisms); in the former example, a typical dimension of the observation vector is of the order of thousands or tens of thousands, while in the latter it is of the order of hundreds of thousands.

In the talk, an up-to-date, albeit brief, overview of the area of supervised learning for small-n-large-p-problems will be provided. While the emphasis will be on classification, problems with quantitative responses will also be dealt with. Issues of model selection and regularization of methods will be presented. Special focus will be placed on the issue of feature selection.

Complex systems of any sorts are characterised by autonomous components interacting with each other in a non-trivial way. In this paper we discuss how the views on complexity are evolving in fields like physics, social sciences, and computer science, and – most significantly – how they are converging. In particular, we focus on the role of interaction as the most important dimension for modelling complexity, and discuss first how coordination via mediated interaction could determine the general dynamics of complex software system, then how this applies to complex socio-technical systems like social networks.

The problem of stereovision-based on-board perception for autonomous mobile systems is complex and imposes challenging constraints on system hardware and software architecture: a wide field of view of the stereo ensemble, high frame rate, dense and accurate data acquisition and processing, robust object detection, tracking and classification.

The solution researched and developed at Technical University of Cluj-Napoca in the Image Processing and Pattern Recognition Center of Computer Science Department is a novel 6D stereovision based perception solution built on two levels architecture and proposing two 3D environment perception methods.

The low level architecture implements the 6D stereovision sensor, providing the primary data for the upper level, consisting of the 3D position and 3D motion information.

Two alternatives are proposed for the high level architecture in charge with the 3D environment description. A structured approach corresponding to a model based detection, tracking and classification of the environment, fit to highly controlled and structured environments. In structured environments, the obstacles are modeled as cuboids, having position, size, orientation and speed.

For the complex or unstructured environments a more general approach called unstructured approach is proposed. The 3D position information provided by the stereovision is used to build a Digital Elevation Map. Elevation map cells are classified into road, obstacles and intermediate generating the Classified Occupancy Maps. Dynamic Occupancy Maps can be generated by associating the 3D motion information to the Classified Occupancy Maps. This set of maps allows the implementation of more effective modeling and inference mechanisms corresponding to the static and dynamic entities from environment.

The main original contributions highlighted in the presentation are the following:

• A new stereo-reconstruction solution using an improved SGM approach (SORT_SGM)
• Original solutions for dense optical estimation
• Ego motion estimation from stereo sequences
• Lane and road geometry perception
• Painted road objects perception
• Obstacles detection, tracking and classification
• Perception and representation of the unstructured environment
• Dynamic Environment Modeling
• Forward collision detection
• High level reasoning on perception and domain knowledge