Spectrum of Rule Algorithms: From Classical, through Multiple-instance, to One-class
Krzysztof (Krys) Cios
Virginia Commonwealth University, USA

 [Abstract]
In the talk I will review supervised inductive machine learning algorithms that generate rules and justify why they are a preferred choice for model building in many domains.
First, I will introduce a classical rule learner that is scalable to big data and that is a building block for subsequently developed other algorithms.
Second, I will introduce multiple-instance learning (MIL) problem and algorithm.
In MIL we do not have unique instances corresponding to unique classes but we have "bags" of instances and we know only that a bag is positive if at least one of its instances is positive, or negative if all of its instances are negative.
Third, I will talk about one-class learning where only one, target class, of instances is available and there is no corresponding class information. This type of learning is also known as novelty/outlier detection problem.

 [Biography]
Krzysztof "Krys" Cios (Ph.D., D.Sc., MBA) is Professor and Chair of Computer Science Department at the Virginia Commonwealth University, U.S.A. His research interests are in the areas of machine learning/data mining and biomedical informatics/computational neuroscience. His research was funded by NIH, NASA, NSF, NATO, and the U.S. Air Force. Prof. Cios served as primary advisor to 15 doctoral students who now work as university professors, post-doctoral researchers, or at companies such as NASA, General Motors, and Proctor and Gamble. He published three books and close to 200 journal and conference papers. Prof. Cios has been the recipient of the Norbert Wiener Outstanding Paper Award, the Neurocomputing Best Paper Award, and the Fulbright Senior Scholar Award. He is a Foreign Member of the Polish Academy of Arts and Sciences.

Hyper-heuristics: Automatic Design of Optimization Algorithms with Grammatical Evolution
Francisco B. Pereira
Instituto Superior de Engenharia de Coimbra, Portugal

 [Abstract]
The goal of hyper-heuristics is to automate the design of heuristic methods used to solve hard optimization problems. Hyper-heuristics operate on the search space of heuristics, seeking for innovative search methods for a given optimization task.
In this talk we present a complete framework to evolve the architecture of a full-fledged Ant Colony Optimization algorithm. Grammatical Evolution is adopted as the design methodology, as it allows for a simple representation and modification of flexible algorithmic strategies. The grammar used by this framework has the ability to guide the learning of novel architectures, by rearranging components regularly found on human designed variants.
Results obtained with the Traveling Salesperson Problem show that the evolved strategies are effective, exhibit a good generalization capability, and are competitive with human designed variants.

 [Biography]
Francisco B. Pereira is a Professor at the Informatics Engineering Department from the Polytechnic Institute of Coimbra, Portugal. He is also a researcher at the Evolutionary and Complex Systems Group, a research line from the Centre for Informatics and Systems of the University of Coimbra, Portugal. His research interests are focused on the development and application of bio-inspired algorithms to combinatorial and numerical optimization problems.
He edited one book on bio-inspired approaches to the vehicle routing problem and published over 60 peer-reviewed scientific papers in journals and conference proceedings. He received three best paper awards at the conferences EuroGP-1999, EvoLearn-2001, and EuroGP-2012.
In the past 15 years he has given several invited talks/lectures in scientific events and academic institutions and organized seven workshops/special sessions in international conferences from the evolutionary computation area. Also, he has been a regular member from the program committee of several international conferences. He is Conference Co-chair of the forthcoming XVII Portuguese Conference on Artificial Intelligence (EPIA-2015) that will take place in Coimbra, in September 2015.

So, What's New About Social Networks ?
Benjamim Fonseca,
University of Trás-os-Montes e Alto Douro , Portugal

 [Abstract]
Strictly, social networks are nothing new, they always existed as groups of people that somehow organize themselves around a specific interest or set of interests. Typical examples are technical societies, political parties, residents’ committees or secret societies. In this perspective, Facebook, Google+ or Twitter are not social networks, they are social media, platforms that provide a set of tools that allows us to build our own social networks – friend’s lists, circles, Facebook groups, etc.
The main new thing about social media is that it provides powerful technological tools that speeds up our social networks construction, maintenance and expansion, allowing easy ways to find people, sharing content instantly and looking at others’ activities. Many of us share and store thoughts, photos and other media in Facebook, Twitter, Flickr or Instagram. At what extent can we say that social media is changing the way we organize ourselves?
An important thing about social media is that many people to whom was not so easy to participate in traditional social networks, such as teenagers, disabled or shy people, found there a way to facilitate its participation or even be active social network promoters.
The massive presence of teenagers in social media is seen by many people as a danger but many others rather prefer to look at it as a big challenge. What can we do with this? How can we take this opportunity to easily reach teenagers’ social networks and thus use social media as an educational tool?

 [Biography]
Benjamim Fonseca is an Assistant Professor at the University of Trás-os-Montes e Alto Douro (UTAD), in Portugal, where he lectures on collaboration, social networks and inclusive systems, and a researcher at the INESC TEC - Associate Laboratory. He received a PhD from UTAD, with a thesis subject of "A model for creating cooperative services". His research interests are Computer Supported Cooperative Work (CSCW), social media and mobile accessibility, having participated in several research projects funded by companies and international funding agencies. He authored or coauthored over 80 scientific publications in conferences, journals and books in these research fields, and actively participates in the organization and scientific committees of several reputed conferences and journals.

Biologically Inspired Algorithms for Speech Based Interfaces
Luis Coelho,
Escola Superior de Estudos Industriais e de Gestão, Portugal

 [Abstract]
Speech is the most natural way of communicating for humans and the dream of creating a machine who is able to mimic such ability has been around for several decades. In 1968, in Kubric's famous Space Odyssey movie, a conversation with a computer (the Hal) was already imagined to be possible in 2001. However the study of speech based interaction still has many open challenges and the quest for algorithms that can deliver abilities that are natural for humans has not yet finished.
This talk will cover the main biologically inspired algorithms and structures used in speech interfaces, what are the main open problems and how future research is envisioned.

 [Biography]
Luis Coelho, PhD, has been doing research in speech science since 2004 and has collaborated in several projects with public and private organizations. His main interests in the area are machine learning and signal processing. He is currently teacher at the Polytechnic Institute of Porto and the coordinator of the Biomedical Engineering Degree.