Human-Agent Teamwork in Cyber Defense
Abstract: Despite the significant attention being given to the critical problems of cyber security, the ability to keep up with the increasing volume and sophistication of network attacks is seriously lagging. Throwing more computing horsepower at fundamentally-limited visualization and analytic approaches will not get us anywhere. Instead, we need to seriously rethink the way cyber security tools and approaches have been conceived, developed, and deployed.
IHMC is taking advantage of the combined strengths of humans and software agents to create new capabilities for Network Operations Centers (NOCs). These capabilities are being implemented in a new cyber defense framework called Sol. Our objective is to enable distributed sensemaking, rapid detection of threats, and effective protection of critical resources. Specifically, we use agents, policies, and visualization to enact coactive emergence as a sensemaking strategy for taskwork and teamwork, and we implement capabilities for organic resilience and semantically-rich policy governance as a means of assuring effective and adaptive human-agent team response.
IHMC has applied its long years of experience with software agents to the design of a new agent framework called Luna. Luna agents function both as interactive assistants to analysts and as continuously-running background aids to data processing and knowledge discovery. Luna agents achieve much of their power through built-in teamwork capabilities that, in conjunction with IHMC's KAoS policy services framework, allow them to be proactive, collaborative, observable, and directable. In order to support dynamic scalability and other features of the Sol framework, the Luna platform supports the policy-governed option of allowing the state of agents (vs. code of agents) to migrate between operating environments and hosts.
We believe that the approach to cyber defense embodied in Sol is equally relevant to applications of distributed sensemaking for other kinds of complex high-tempo tasks such as real-time disease control or disaster management.
Biography: Jeffrey M. Bradshaw (Ph.D., Cognitive Science, University of Washington) is a Senior Research Scientist at the Florida Institute for Human and Machine Cognition (IHMC) where he leads the research group developing the KAoS policy and domain services framework. He co-leads the group developing IHMC's Sol Cyber Framework and the Luna agent framework. Formerly, Jeff led research groups at The Boeing Company and the Fred Hutchinson Cancer Research Center. Though his earliest publications were focused on memory and language, Jeff's research focus soon turned to a wide variety of topics relating human and machine intelligence. With Ken Ford, he edited the seminal volume Knowledge Acquisition as a Modeling Activity, and became well known for his role in helping develop a suite of successful methodologies and tools for automated knowledge acquisition. While at Boeing, he founded the emerging technologies group of the Aviation Industry Computer-Based Training Committee (AICC). Jeff helped pioneer the research area of multi-agent systems, and his first book on the topic, Software Agents, became a classic in the field and a best-seller for The MIT Press. Jeff served for over a decade on the Board of Directors of the International Foundation for Autonomous Agents and Multiagent Systems.
Human-Agent-Robot Teamwork has been a central interest for many years. From 2002-2006, KAoS was used as part of a NASA series of annual two-week field tests of human-robot teams performing simulated planetary surface exploration at the Mars Desert Research Station in the Utah desert. Jeff was sponsored by DHS to undertake detailed simulation studies of the use of human-robot teams to secure facilities at Port Everglades. He has also led the ONR-sponsored NAIMT and Coordinated Operations projects where a team of humans and heterogeneous robots performed field exercises at the Naval Air Station in Pensacola, aimed at port reconnaissance, and robot-assisted detection and apprehension of intruders. Jeff co-founded and organized the Human-Agent-Robot Teamwork Workshop series (HART), whose most recent meetings were held at the Lorentz Center in Leiden, The Netherlands, and at the 2012 Human-Robot Interaction conference in Boston. He recently served as lead editor for a special issue of IEEE Intelligent Systems on HART (April-May 2012). In June 2012, he led an international workshop for the National Academies on Intelligent Human-Machine Collaboration.
Jeff has been a Fulbright Senior Scholar at the European Institute for Cognitive Sciences and Engineering (EURISCO) in Toulouse, France; a visiting professor at the Institut Cognitique at the University of Bordeaux; is former chair of ACM SIGART; and former chair of the RIACS Science Council for NASA Ames Research Center. He served as a member of the National Research Council (NRC) Committee on Emerging Cognitive Neuroscience Research, was an advisor to the HCI and Visualization program at the German National AI Research Center (DFKI), and was a scientific advisor to the Japanese NEC Technology Paradigm Shifts initiative. He currently serves as a member of the Board on Global Science and Technology for the National Academies and as an external advisory board member of the Cognitive Science and Technology Program at Sandia National Laboratories. He is an Honorary Visiting Researcher at the University of Edinburgh, Scotland, a member of the Graduate Faculty at the Florida Institute of Technology, a faculty associate at the University of West Florida. With Robert Hoffman and Ken Ford, he serves as co-editor of the Human-Centered Computing Department for IEEE Intelligent Systems. In 2011, he received the Web Intelligence Consortium Outstanding Contributions Award.
25 Years of Distributed AI in Germany: From the very beginnings to multiagent applications in industry in Germany today
Abstract: I. The presentation starts with a brief report on the first steps of Distributed AI research in Germany, beginning in 1988, the year of the publication of the famous book of Bond / Gasser, followed by a stimulating discussion on Michael Huhns DAI mailing list (1991), and by a number of ambitious German PhD projects (Fischer, Klett, v. Martial, Kirn and others), and, finally, by the foundation of the German Special Interest Group on Distributed AI in the early 1990´s.
II. From its very beginning, German DAI research has been driven by three main forces:
(1) basic research: developing towards well-defined concepts for distributed problem solving and multiagent systems, coordination and cooperation, situated behaviors, etc. in two 6 years basic research programs on Socionik (1999-2005), and cooperative intelligent agents (2000-2006).
(2) industrial DAI/MAS applications in lots of branches, e.g. banking and finance, production planning and control, logistics and others, and
(3) the enrichment of other computer science technologies like databases, knowledge based systems, distributed systems, grid computing, and cloud computing nowadays by proven concepts of the multiagent field.
This part of the presentation gives introdcutions into and examples of these projects and to the results achieved.
III. Nowadays it is well understood that the multiagent paradigm is an important cutting edge technology for innovative solutions in the field of smart objects. National and international research programs offer a lot of opportunities for ambitious multiagent research and development. The presentation introduces some up to date examples from fields like food security, material logistics, and civil engineering.
Biography: Prof. Dr. Stefan Kirn holds a chair in Information Systems at University of Hohenheim at Stuttgart/Germany. He obtained master degrees in management science (Munich, 1980) and in computer science (Hagen, 1989). He did his Ph.D. in Distributed AI in 1991, and his habilitation in Organizational Theories for Multi-Agent Systems in 1995. Stefan Kirn has been the initiator and coordinator of a 6 years German Research Foundation priority program on business applications of agent technology (2000-2006). From this research program originated the agent technology testbed Agent.Hospital, which has been successfully presented to an international audience the first time at Barcelona in February 2003. The most important results of this program have been published as a Springer book entitled "Multiagent Engineering Handbok". From 2001-2003 he has been Chief Scientist of IWT GmbH/Erfurt, a research company owned by the Association of the Industry of Thuringia/Germany. Since 2005, Stefan Kirn is CEO of the Hohenheim Research Center on Innovation and Services. In 2004, he founded Jesselle GmbH, an agent technology based university spinoff. Since 2004, he initiated a series of series of national and European projects on industrial applications of multiagent technology, e.g. Akogrimo, AutoBauLog, BREIN, MIGRATE!, ProBauDok, etc. Stefan Kirn has published more than hundred peer reviewed international publications, is on the editorial board of several journals in information systems, and acts as a consultant to governments, industry, and professional organizations.