Keynote I: Belief functions: past, present and future

     The theory of belief functions, sometimes referred to as evidence theory or Dempster-Shafer theory, was first introduced by Arthur P. Dempster in the context of statistical inference, to be later developed by Glenn Shafer as a general framework for modelling epistemic uncertainty. Belief theory and the closely related random set theory form a natural framework for modelling situations in which data are missing or scarce: think of extremely rare events such as volcanic eruptions or power plant meltdowns, problems subject to huge uncertainties due to the number and complexity of the factors involved (e.g. climate change), but also the all-important issue with generalisation from small training sets in machine learning. This keynote talk, abstracted from a recent half-day tutorial at IJCAI 2016, is designed to introduce to non-experts the principles and rationale of random sets and belief function theory, review its rationale in the context of frequentist and Bayesian interpretations of probability but also in relationship with the other main approaches to non-additive probability, survey the key elements of the methodology and the most recent developments, discuss current trends in both its theory and applications. Finally, a research program for the future is outlined, which include a robustification of Vapnik' statistical learning theory for an Artificial Intelligence 'in the wild'.

  Keynote II: Wireless Sensor Networks and energy harvesting: Opportunities, Challenges, and Research Trends

     The growing gap between the complexity of today's motes and the capacity of their batteries makes power management strategies in WSN insufficient and rises the need to explore innovative solutions. Energy harvesting from the environmental resources (e.g., solar, wind, electro magnetic, thermal, etc.) is appearing to be the key solution, as the appropriate technologies are constantly evolving and maturing, and their complexity is expected to be reduced in the near future to provide adequate components for WSN. Augmenting the sensor motes with the ambient energy harvesting capability will eliminate the problem of permanent battery drying out (dead nodes) and enable large scale deployment of sustainable wireless sensor networks in the future. However, satisfying communication requirements with the available energy profile is a nontrivial problem that strongly depends on the EH technology. Existing models and architectures should be revisited and rebuilt upon an energy model that exploits this new feature, while faithfully reflecting the real world constraints for harvesting. This talk introduces energy harvesting wireless sensor networks and discusses some related issues. Some relevant results from the CERIST's WSN research group will also be sketched.

  Keynote III: Exploring Classical Artificial Intelligence, Deep Machine Learning and Biological Neural Network Learning

     We start by introducing the main features of classical artificial intelligence and its various domains of application. The pros and cons of this paradigm will be discussed. We follow up by covering the artificial neural network approach to intelligence as well as its backpropagation learning scheme. The conditions for the exploitation of these machines will be outlined through some selected examples. Deep machine learning schemes including "Convolutional Neural Networks" (CNN) as an extension to the conventional neural netsbased machine learning will also be set forth. Likewise, the advantages and limitations of these learning models will be laid out further. Finally, we address the biological neural networks approach such as the “Hierarchical Temporal Memories” (HTM) and the concept of "Sparse Distributed Representation" (SDR) to mimic the human brain’s structure considered by many researchers in the field as the roadmap for creating future intelligent machines. We conclude the speech by exhibiting a problem in which data are constantly changing and show how these three formalisms of artificial intelligence address it.

  Keynote IV: Drawing Parallels between Nature and Cloud Computing for Enhanced Resilience and Survivability

     As Organisations are increasingly adopting cloud-computing as the foundation for their IT infrastructure, the reliability of inherently complex cloud systems becomes under test. In cloud computing, the robustness of the infrastructure and service and the overall survivability is enhanced by creating redundancy for backup in times of fault, failure or attack. Concepts and processes existing as nature's inherently multifunctional capabilities such as robustness, resiliency, survivability, and adaptability, could provide inspiration for unconventional methods to solve unique problems in the computing continuum. Ensuring the resilience of critical infrastructures is ever more necessary with the increasing threat of cyber-attacks, due to the increased complexity. It is generally acceptable that whilst complexity increases resilience and reliability decreases. However Biological systems subvert this rule; they are inherently much more complex, yet highly reliable. The strengths inherent in biological systems resides in the ability of autonomous entities to make local decisions, continuously coordinate and share information, while maintaining a global form of order. Furthermore, the challenges of attaining survivability have also been successfully addressed by nature, and effectively demonstrated in attributes of collaborative communities. This keynote will draw parallels between capabilities in nature such as those demonstrated in multi-cellular biological systems and by the predation avoidance in primate in predator-prey communities, and capabilities in security systems in cloud environments.