ICCM 2015 will host two tutorials and one workshop on the day before the actual conference (Wednesday, April 8). We believe that these high-quality tutorials and workshop will be of interest to many of the conference attendees. You can register for one of these during the general ICCM registration, they’re included in the conference fee.
Eduardo Alonso – School of Mathematics, Computer Science and Engineering, City University London, UK
Esther Mondragon – Centre for Computational and Animal Learning Research Ltd., St Albans, UK
(half-day tutorial)
It is widely accepted that classical conditioning is at the basis of most learning phenomena and behavior and thus paramount that we understand well how its mechanisms have been modeled computationally. This tutorial will focus on two influential implementations, namely Rescorla and Wagner’s model (RW) and Temporal Difference (TD, aka Reinforcement Learning). In the first hour we will present the fundamentals of classical conditioning and error-correction learning, and explain in more detail how these are instantiated in both trial-based (RW) and real-time (TD) representations. The next two hours will be dedicated to run simulations of the two models in specialized software, RW Simulator and TD Simulator respectively. The tutees will learn the basics of their functioning, how to input data (experimental designs, parameter values) and how to interpret output against a selected collection of experimental data as well as with regards to novel predictions.
Terry Stewart & Peter Blouw
University of Waterloo, Centre for Theoretical Neuroscience
(full-day tutorial)
This tutorial introduces the Neural Engineering Framework, a general-purpose method for implementing computational algorithms using realistic spiking neurons. It has been used to develop biologically realistic models of visual processing, motor control, planning, mental arithmetic, and analogy, as well as Spaun, the first large-scale brain simulation capable of performing multiple tasks. In this tutorial, we cover the underlying theoretical framework while giving hands-on examples of building and running models using Nengo, our cross-platform open-source neural simulation software. Participants will be expected to bring a laptop and follow along.
While we have provided this tutorial in the past, there are two major differences this year. First, the newest version of Nengo has extensive support for running on different hardware such as GPUs, allowing for significantly larger neural models. Secondly, these larger models allow for more complex cognitive models, including novel approaches to linguistic processing and complex rule following.
John Laird – University of Michigan
Kevin Gluck – U.S. Air Force Research Laboratory
(full-day workshop)
Advances in Artificial Intelligence, Cognitive Science, and Robotics are leading us to future autonomous systems that will have the cognitive and physical capabilities to perform a wide variety of tasks, with applications across science, health care, business, military, home, and entertainment. How will these robotic and agent systems learn the unanticipated and evolving complex tasks we want them to perform? Although many related areas of science and technology will play a role in answering this question, the focus of this workshop is on how cognitive modeling research can inform the quest for achieving interactive task learning – agents and robots acquiring new tasks through natural interactions with humans. What capabilities can we identify and model in humans that enable (and are potentially necessary for) interactive task learning, what are the most important relevant successes so far, and what are the most important research challenges that need to be addressed?
We seek a balance between thought-provoking keynote-style presentations and active participation by workshop attendees in producing answers to the three questions mentioned above in the abstract. Our initial schedule will be:
0900 – Introduction: John Laird & Kevin Gluck
0930 – Invited Presentation: Michael Beetz (Robotics; University of Bremen)
1030 – Break
1045 – Discussion Groups (Cognitive Capabilities)
1200 – Brief Group Reports
1230 – Lunch
1400 – Invited Presentation: Ken Forbus (Artificial Intelligence; Northwestern University)
1500 – Break
1515 – Discussion Groups (Successes; Challenges)
1630 – Brief Group Reports
