Invited speakers
Hod Lipson
The Robotic Scientist: Mining experimental data for dynamical invariants, from cognitive robotics to computational biology
ABSTRACT: For centuries, scientists have attempted to identify and document analytical laws that underlie physical phenomena in nature. Despite the prevalence of computing power, the process of finding natural laws and their corresponding equations has resisted automation. A key challenge to finding analytic relations automatically is defining algorithmically what makes a correlation in observed data important and insightful. By seeking dynamical invariants, we go from finding just predictive models to finding deeper conservation laws. We demonstrated this approach by automatically searching motion-tracking data captured from various physical systems, ranging from simple harmonic oscillators to chaotic double-pendula. Without any prior knowledge about physics, kinematics, or geometry, the algorithm discovered Hamiltonians, Lagrangians, and other laws of geometric and momentum conservation. The discovery rate accelerated as laws found for simpler systems were used to bootstrap explanations for more complex systems, gradually uncovering the "alphabet" used to describe those systems. Application to modeling physical and biological systems will be shown.
BIO: Hod Lipson is an
Associate Professor of Mechanical & Aerospace Engineering and Computing &
Information Science at Cornell University in Ithaca, NY. He directs the
Computational Synthesis group, which focuses on novel ways for automatic
design, fabrication and adaptation of virtual and physical machines. He has
led work in areas such as evolutionary robotics, multi-material functional
rapid prototyping, machine self-replication and programmable
self-assembly. Lipson received his Ph.D. from the Technion - Israel
Institute of Technology in 1998, and continued to a postdoc at Brandeis
University and MIT. His research focuses primarily on biologically-inspired
approaches, as they bring new ideas to engineering and new engineering
insights into biology. For more information visit http://www.mae.cornell.edu/lipson.
Marie-Francine Moens
More than Just Words: Discovering the Semantics of Text with a Minimum of Supervision
ABSTRACT: Humans discover many different aspects of meaning in sources such as text, speech, audio and visual data. We retrieve certain information from them, which is used in our daily actions and decisions. Machines can assist in these tasks when they mine, summarize or aggregate content from the digital sources. We can train pattern recognizers to discover the semantics of the sources, but a major bottleneck is the lack of sufficient annotated examples, as a manual labeling is often prohibitively expensive. A multitude of semantic classes can be recognized from the surface features and the variation of surface features that express a similar meaning is usually high, making the task not easier.
In this talk we explore the possibilities of reducing training examples when recognizing semantics in text. We discuss approaches of unsupervised expansion of the training set, multiple instance learning, active learning and unsupervised language modeling, and study their impact on semantic classification of text. We illustrate the approaches by our own research on relationship detection, semantic role labeling, opinion extraction and argumentation mining. We demonstrate that the large amounts of data that can be cheaply collected (e.g., on the World Wide Web) assist in recognizing the targeted information. Linguistic and cognitive theories can guide us in the selection of appropriate features and seed training examples.
BIO: Marie-Francine Moens is associate
professor at the Department of Computer Science of the Katholieke
Universiteit Leuven, Belgium. She holds a Ph.D. degree in Computer Science
(1999) from this university. She currently leads a research team of 1
postdoctoral fellow and 9 doctoral students, and is currently coordinator
of or partner in 7 European or international research projects in the
fields of information retrieval and text mining. Her main interests are in
the domain of automated content retrieval from texts with a strong emphasis
on probabilistic content models obtained through machine learning
techniques. Since 2001 she teaches the course Text Based Information
Retrieval and since 2009 she partly teaches the courses Natural Language
Processing and Current Trends in Databases at K.U.Leuven. She has
(co-)authored more than 150 research papers in the field of information
retrieval and text analysis, is author of two monographs published in the
Springer International Series on Information Retrieval, and is (co-)editor
of several books. She is the (co-)organizer of 3 editions of the KRAQ
Knowledge and Reasoning for Answering Questions conferences (respectively
at IJCAI 2005, COLING 2008 and ACL 2009), of the ACM SIGKDD Workshop on
Cybersecurity and Intelligence Informatics (KDD 2009), and the Cross-media
Information Access and Mining workshop (IJCAI-AAAI 2009). She is currently
appointed as chair-elect of the European Chapter of the Association for
Computational Linguistics (2009-2010).
http://www.cs.kuleuven.be/~sien/
Associate Professor
Department of Computer Science
Katholieke Universiteit Leuven
Belgium
Demetri Terzopoulos
Artificial Life Simulation of Humans and Lower Animals: From Biomechanics to Intelligence
ABSTRACT: The confluence of virtual reality and artificial life, an emerging discipline that spans the computational and biological sciences, has yielded synthetic worlds inhabited by realistic artificial flora and fauna. The latter are complex synthetic organisms with functional, biomechanically-simulated bodies, sensors, and brains with locomotion, perception, behavior, learning, and cognition centers. These biomimetic autonomous agents in their realistic virtual worlds foster deeper computationally-oriented insights into natural living systems. Virtual humans and lower animals are of great interest in computer graphics because they are self-animating graphical characters poised to dramatically advance the motion picture and interactive game industries. Furthermore, they engender interesting new applications in computer vision, medical imaging, sensor networks, archaeology, and many other domains.
BIO: Demetri Terzopoulos (PhD '84 MIT) is the
Chancellor's Professor of Computer Science at UCLA. He is a Guggenheim
Fellow, a Fellow of the ACM, a Fellow of the IEEE, a Fellow of the Royal
Society of Canada, and a member of the European Academy of Sciences. One of
the most highly cited authors in engineering and computer science, his
numerous awards include an Academy Award for Technical Achievement from the
Academy of Motion Picture Arts and Sciences for his pioneering research on
physics-based computer animation, and the inaugural Computer Vision
Significant Researcher Award from the IEEE for his pioneering and sustained
research on deformable models and their applications.
http://www.cs.ucla.edu/~dt/
The Chancellor's Professor of Computer Science
University of California, Los Angeles
Henry Samueli School of Engineering and Applied Science
Computer Science Department
4531G Boelter Hall
Los Angeles, CA 90095-1596, USA