AISB Convention 2015

The AISB Convention is an annual conference covering the range of AI and Cognitive Science, organised by the Society for the Study of Artificial Intelligence and Simulation of Behaviour. The 2015 Convention will be held at the Uni...


Read More...

Yasemin Erden on BBC

AISB Committee member, and Philosophy Programme Director and Lecturer, Dr Yasemin J. Erden interviewed for the BBC on 29 October 2013. Speaking on the Today programme for BBC Radio 4, as well as the Business Report for BBC world N...


Read More...

Mark Bishop on BBC ...

Mark Bishop, Chair of the Study of Artificial Intelligence and the Simulation of Behaviour, appeared on Newsnight to discuss the ethics of ‘killer robots’. He was approached to give his view on a report raising questions on the et...


Read More...

AISB YouTube Channel

The AISB has launched a YouTube channel: http://www.youtube.com/user/AISBTube (http://www.youtube.com/user/AISBTube). The channel currently holds a number of videos from the AISB 2010 Convention. Videos include the AISB round t...


Read More...

Lighthill Debates

The Lighthill debates from 1973 are now available on YouTube. You need to a flashplayer enabled browser to view this YouTube video  


Read More...
01234

Notice

AISB miscellaneous Bulletin Item

CFP: IEEE Trans SMC-B Special Issue on ADP and RL for Control

http://www.eecs.wsu.edu/~cook/smcb

IEEE Trans SMC-B Special Issue on Approximate Dynamic Programming and Reinforcement Learning for Control

New deadline for paper submission: September 1, 2007

Neural networks and fuzzy logic systems possess highly attractive mathematical
and adaptive learning properties that have been exploited in numerous fields
including signal processing, sensor fusion, decision-making and
classification, pattern recognition and clustering, multimedia processing, and
elsewhere.

Man-made feedback controllers are responsible for many of the recent
developments in engineered systems including aerospace, ship, vehicle,
industrial, robotic, and process control systems.  Applications of neural
networks and fuzzy logic in feedback control systems are intriguing because
they potentially allow the control of complex engineered systems whose
dynamics are unknown a priori.

Most applications of neural networks for feedback purposes explored by the
Control Systems Community have been to extend the capabilities of controllers
designed via traditional tools by augmenting them with neural networks
employed as nonlinear function approximators for unknown system dynamics.  So
far, these applications have not focused on optimal or efficient control.

In nature, biological and ecological systems operate on various principles of
optimality, efficiency, or economy, based on constrained resources.  It is
known that the solution of various optimal and inverse optimal controller
design equations can potentially provide optimal controllers for any form of
nonlinear system.  Problems related to determining a control input that
minimizes various prescribed cost or value functionals can be confronted by
solving so-called Hamilton-Jacobi-Bellman equations.  These HJB equations are
generally nonlinear and impossible to solve analytically.

A large and highly successful body of work exists in reinforcement learning
and Adaptive Dynamic Programming (ADP), also called Neurodynamic Programming
(NDP), which approximately solves the HJB equations using efficient
forward-in-time techniques based on using neural networks for value functional
approximation.  Much work has been done in this area to develop various ADP
algorithms, prove convergence, determine effective solution procedures, and
design practical control systems for aerospace systems and elsewhere.

ADP holds out the promise of providing a new class of Adaptive Controllers
that directly converge to optimal control designs.  

The purpose of this Special Issue is to showcase the state-of-the-art in ADP,
NDP, and reinforcement learning for feedback control applications.

Topics of Coverage: Including but not limited to
- Adaptive dynamic programming
- Reinforcement learning for system control
- Approximately optimal control
- Value function approximation for control systems design
- Q learning (action-dependent learning)
- Neural network applications in dynamic programming for feedback systems
- Fuzzy logic systems and fuzzy-neural systems for dynamic programming
- Direct policy search
- Actor-critic methods
- Learning rules and architectures for ADP
- Partially observable Markov decision processes
- Approximate solutions to nonlinear controller design equations
- Approximation-based optimal control systems
- Applications of ADP

Guest Editors
F.L. Lewis, Univ. of Texas at Arlington, Texas
George G. Lendaris, Portland State University, Portland, Oregon
Derong Liu, University of Illinois at Chicago

Submission
Papers for the special issue must be submitted electronically through
Manuscript Central at http://smcb-ieee.manuscriptcentral.com. 
Information regarding electronic submission can be found at
http://www.eecs.wsu.edu/~cook/smcb.
Each submitted paper must indicate that the paper is submitted to the special
issue on ADP.  Submitted manuscripts will be screened for topical relevance,
and those relevant to the special issue will undergo IEEE SMC-B standard
review procedures.

The paper submission deadline is: 1st September 2007