We have invited a distinguished set of invited speakers to the Cognitive Systems conference. Confirmed speakers include

Michael L. Anderson

Chair, Scientific and Philosophical Studies of Mind Program
Associate Professor, Department of Psychology
Franklin & Marshall College
Lancaster, PA 17604
(http://www.fandm.edu/michael-anderson, http://www.agcognition.org)

Using structure where you find it: Symbol systems in multi-step cognition

Abstract: This talk will outline the important role of overt action and environmental interaction in higher-order human cognition. I will offer a situated account of Turing computation, and review some of the evidence for the importance of interaction with external symbols in multi-step cognition, mathematics in particular. I will end with a review of the various roles that language plays in structuring complex cognition, and discuss possible implications of these findings for designers of intelligent devices.

Yolanda Gil

Information Sciences Institute and Department of Computer Science
University of Southern California
Dr. Yolanda Gil is Director of Knowledge Technologies and Associate Division Director at the Information Sciences Institute of the University of Southern California, and Research Professor in the Computer Science Department. She received her M.S. and Ph. D. degrees in Computer Science from Carnegie Mellon University. Dr. Gil conducts research on various aspects of interactive knowledge capture, including intelligent user interfaces, knowledge-rich problem solving, and the semantic web. In recent years, her work has focused on collaborative large-scale data analysis through semantic workflows. She initiated and chaired the W3C Provenance Incubator that led to the PROV standard. Dr. Gil is the current Chair of ACM SIGAI, the Association for Computing Machinery's Special Interest Group on Artificial Intelligence. She was elected Fellow of the American Association of Artificial Intelligence (AAAI) in 2012.

The Human Bottleneck in Data Analytics: Opportunities for Cognitive Systems in Automating Scientific Discovery

Abstract: While orders of magnitude improvements in computing, network bandwidth, and distributed sensing are pushing the envelope in the scale of the scientific phenomena that can be studied, the human component of data analytics has been largely unaddressed and is increasingly becoming a bottleneck to progress. This has created great opportunities for artificial intelligence to capture human expertise and develop knowledge-driven infrastructure that can make data analysis processes more efficient and can break new barriers in the complexity of the problems that can be tackled. In this talk, I will describe our current research on intelligent workflow systems that provide assistance and automation for complex data analysis processes. Workflows capture useful combinations of analytic tools and their dataflow as reusable multi-step methods. We have extended workflow representations with semantic constraints that express characteristics of data and analytic models. We have developed algorithms that use those constraints to automatically explore the space of possible analyses given high-level user guidance. Our WINGS workflow system implements these algorithms on top of semantic web standards and reasoners. Data analysis experts can use semantic workflow representations in WINGS to define constraints about the use of analytic tools and their data requirements. End users can get intelligent assistance from WINGS to validate and easily run workflows with their own datasets. In more recent work, we are addressing the challenge of collaborative creation of workflows in a framework that we call "organic data science", where the expertise of diverse scientists is brought in as the need for specific data, models, skills, or knowledge is uncovered as the collaboration progresses. The development of intelligent frameworks for scientific data analysis will enable scientists to tackle increasingly more challenging integrative problems in a more cost-effective and therefore more ubiquitous manner.

Patrick Winston

Patrick Winston is Ford Professor of Artificial Intelligence and MacVicar Faculty Fellow at the Massachusetts Institute of Technology. A graduate of MIT, his Genesis research group focuses on developing a computational account of human intelligence and how it differs from that of other species, with special attention to modeling human story comprehension, telling, and composition. He is now Research Coordinator for the multiuniversity, multdisciplinary Center for Brains, Minds, and Machines centered at MIT.

Getting beyond the shadows on the Wall

Abstract: These are exciting times. IBM's Jeopardy playing system and Google's picture-captioning demonstrations fire the imagination and demonstrate that, with enough data, a machine-learning system can exhibit a measure of intelligence without being very cognitive. Such systems, engineering marvels to be sure, are nevertheless like the chained-up, shadow-watching prisoners in Plato's Republic: they have no connection to reality as we know it and they do not know in any serious sense how they do what they do.

I argue that if we are to go beyond the mere appearance of intelligence, then we have to learn to build systems that understand, tell, and create stories. Such systems can exhibit complex concept composition, human-like reasoning, and a kind of creativity.

To illustrate, I will explain aspects of Genesis System, built by my students and me. Genesis embodies a model of aspects of human story processing so as to read simple stories, answer questions about them, exhibit cultural bias in interpretation, develop trait-driven expectations, retrieve precedents using concepts, teach instructively, and tell persuasively.