Invited Talks

1. Trust Nothing. Run Encrypted. The Emergence of Tamper-Proof Secure Cloud Computing

Radu Sion
Professor at Stony Brook University, the Director of the National Security Institute, and the CEO of Private Machines Inc.

Abstract: Cloud Computing promises tremendous benefits including reduced costs, higher security, increased agility, and significant scalability. However once clients move to the cloud they lose control. Data and workloads are managed by somebody else and run in a foreign environment. Existing network isolation and standard encryption are simply not enough to protect workloads, because, once active, workloads, keys, and data are fully exposed. As a result security-conscious enterprises keep highly-sensitive workloads on-prem. Such hybrid setups create non-trivial architectural challenges and bottlenecks, impacting overall deployment security, performance, and cost. In this talk we introduce the world’s first tamper-proof cloud technology that changes all that. It provides tamper-proof private compute instances wherein all processing is physically and logically protected. Clients do not have to trust the cloud operator, even with physical access to the hardware. Clients can remotely verify security and computation privacy guarantees, while also attesting to the certified firmware and software stacks. Unauthorised parties or the cloud itself cannot access running computation and data or leak it to third parties, even with physical access and even when compelled. The cloud becomes a secure integral part of client infrastructure, accessible to no-one else, protected not only at rest and in transit but also at runtime during processing. We will demonstrate this technology in OpenStack and discuss how to enable your OpenStack deployment to provide secure instances.

Bio: Radu Sion is a Professor at Stony Brook University, the Director of the National Security Institute, and the CEO of Private Machines Inc. Radu’s research is in Cyber Security and Large Scale Computing. He has published 100+ peer reviewed works in top venues, and has organized 65+ conferences. Dr. Sion has received the National Science Foundation CAREER award for his work on cloud computing security. Radu has worked with and received funding from numerous industry and government partners, including the US Air Force, the Office of the Secretary of Defense, the Department of Homeland Security, the US Army, the Intelligence ARPA, the Office of Naval Research, Northrop Grumman, IBM, NOKIA, Motorola, Xerox Parc, Microsoft, SAP, CA Technologies, the National Science Foundation, and many others. Radu is currently leading Private Machines Inc, a cyber security startup designing the next generation secure cloud computing technologies.

2. Decision Support Systems in Water Management: Challenges and Future Prospects

Andreja Jonoski
Associate Professor of Hydroinformatics, UNESCO-IHE Institute for water education, Delft, the Netherlands

Abstract: Decision support systems (DSSs) for managing natural or man-made water systems presents many challenges, both technical and social. The complexity of the physical systems in question requires adequate integration of monitoring, modelling and decision support components, involving latest advances of sensing technologies, complementary use of physically-based and data-driven (machine learning) models, and application of decision support methods such as optimization and multi criteria analyses. Technical challenges associated with this integration are primarily efficiency-driven, such as standards for sharing water-related data, optimal model-data integration and efficient use of computing resources for model calibration, sensitivity and uncertainty analysis. Given that water is one of the most shared natural resources, decision making in this domain requires taking into account multiple interests, increasingly through participation and collaboration of relevant stakeholders, including citizens. To enable such processes, DSSs in the water domain need to meet additional challenging requirements coming from this complex social context, such as developing of shared understanding among the stakeholders, quantification of trade-offs between conflicting objectives and collaboration support. Demonstration of this sociotechnical nature of decision support in the water domain will be presented, through real world DSSs and examples from recent research projects. Future directions of water-related DSSs will be presented, together with their overlaps with the important initiatives of citizen science and citizen observatories of water and environmental systems.

Bio: Andreja Jonoski is an Associate Professor of Hydroinformatics at the UNESCO-IHE Institute for water education, in Delft, the Netherlands. He works in the field of hydroinformatics for more than two decades, primarily in the areas of modelling different water systems (groundwater aquifers, catchments, urban water supply systems) and their coupling with optimization algorithms for purposes of decision support. His research has been particularly focused on using the Internet and mobile phone networks as platforms for new kinds of decision support systems in the water domain, aimed at broad stakeholder participation. He has participated in more than 10 large European collaborative research projects related to the field of ICT and water management, involving research partners as well as water and ICT industry partners. He has also participated in numerous hydroinformatics research, educational and capacity development projects with partners from Asia, Africa and Latin America.

3. Simulation, Computer Aided Decision-making, and The New Information and Communication Technologies

Acad. Florian Filip
The Romanian National Academy of Sciences

Abstract: The models are used whenever one aims at understanding the properties of a certain existing entity (an object, a situation or a process) or intends to prepare a future action with a view to changing the state of affairs, either through creating a new artifact (by applying design decisions) or through influencing the evolution of a process (by implementing management and control decisions). There are numerous cases when analytic models, which faithfully represent the relationships between input decision variables and system outputs, cannot be set or be effectively handled with available computerized solvers. In a quite large number of practical situations, the decision-maker is eager to make the computer evaluate his/her own decisions which are based on his/her previous experience and acquired knowledge. In such situations, one can resort to simulation models and corresponding software tools. In particular, during the choice phase of H. Simon’s process model of the decision-making activities, “What if?” evaluations of possible coursed pf actions and sensitivity analysis of the chosen solutions are frequently encountered. Simulation is also useful and used in selecting an alternative out of a set of possible courses of actions. The paper is mean to give a balanced presentation of several well established and new methodological issues concerning the usage of computerized simulation models in management and control decision-making processes. A particular emphasis is put on computer supported collaborative activities. The impact of the modern information and communication technologies (I&CT), such as business intelligence and analytics (BI&A), and mobile cloud computing, on the usage of simulation in decision-making activities is discussed too.

Bio: Florin Gheorghe FILIP took his MSc and PhD in control engineering from the TU “Politehnica” of Bucharest in 1970 and 1982, respectively. In 1991, he was elected as a member of the Romanian Academy-RA (The Romanian National Academy of Sciences: www.acad.ro) ). He was a scientific researcher and managing director (1991-1997) of the National R&D Institute in Informatics (ICI) of Bucharest. Currently he is the director of the Library of the Academy. He was elected as vice-president of RA in 2000 and reelected in 2002 and 2006. He was elected, in 2009 and reelected in 2015, as the chair of “Information Science and Technology” section of the RA. His main scientific interests include large–scale systems control, simulation, and optimization, decision support systems, technology management and foresight, application of the IT in the cultural sector.. He is the author/coauthor of about 300 technical and the author/coauthor of 12 monographs and edited/ coedited 24 contribution volumes. More information about him can be found at http://www.academiaromana.ro/sectii/sectia14_informatica/sti_FFilip.htm.

4. Invited talk: Artificial Intelligence and Machine Learning – The Unreasonable Effectiveness of Data

Dan Cautis
Georgetown University

Abstract: Artificial intelligence (AI) is everywhere around us: from the multitude of intelligent robots, state of the art software that translates between languages, recognizes faces, songs and speech to the smart phones that predict typing, deploy advanced GPS, provide fraud detection and to a large number of commercial applications in medicine, traffic control, self driving vehicles, etc. the AI brings a bonanza of benefits in our everyday life.
From the advent of digital computers some half a century ago scientists daringly predicted that “soon” they would be able to develop software that will equal brain’s capabilities so that advanced computers will display human-like intelligence. Those early hopes turned out to be premature since computers had neither enough resources (memory and speed) nor the right software approach for these challenging tasks. After the initial enthusiasm a more pessimistic mood of the research community acknowledged some “AI winters” in the 1970s and late 80s triggered mostly be the collapse of the cumbersome and brittle expert systems that lead to unwieldy, complex, hard to maintain applications which, although useful in some areas, did not seem to have any of the flexible and adaptive characteristics of the human brain. Fortunately, the new field of machine learning came to the rescue and renewed the original optimistic hopes. Remarkable new developments in supervised learning (data pattern recognition, classification, linear and logistic regression), unsupervised learning (clustering) and in particular reinforcement learning (artificial neural networks and deep learning) and especially the recognition of the primary role of data and the statistical nature of learning (Bayesian estimators) helped in creating powerful new approaches that imitate the neuro-cognition processes of the brain. These developments brought to fruition new and very successful applications, significant new investment funding in the field of AI and renewed hopes that creating software that could approach human intelligence could again be placed on the horizon of the foreseeable future.

Bio: Dan Cautis was employed by Georgetown University between 2006 and 2015 as Associate Vice President at Georgetown University. Within UIS (University Information Services) as Director of IT Infrastructure Operations he was responsible for the university’s Data Centers, Computer Operations, Network Operations, Application Production Services, Change Management and Disaster Recovery. Currently, within the Georgetown University SCS, he teaches, as adjunct professor, the following courses: Church and Science – A Historical Perspective where he analyzes the complex philosophical and political interaction between the scientific developments and the Catholic Church dogma from the High Middle Ages until the present time. Transhumanism – Ethical, Social and Religious Implications where he teaches about the scientific and philosophical background of state of the art developments like Technological Singularity, Artificial Super Intelligence, Nanotechnologies and analyzes the ideas and the lively debates within the “Transhumanist Community” that believes in the applications of these technologies to develop a “posthuman” being with better and longer life and superior intelligence. As visiting professor and invited corporate speaker Dan also teaches courses in Artificial Intelligence and Machine Learning using Python implementation of algorithms (with focus on artificial neural networks, data pattern recognition and classification, linear regression, deep learning and reinforcement learning).