Log in

Login to your account

Username *
Password *
Remember Me
From Brain Science to Intelligent Machines

Experience-based goal generation and motivated reinforcement learning for a mobile robot

Date: Thursday 23/01/2014
Venue: MS020
Time: 1.00 pm
Speaker: Dr Kathryn E. Merrick
Affiliation: University of New South Wales, Australian Defence Force Academy, Canberra, ACT, Australia


Experience-based goal generation and motivated reinforcement learning for a mobile robot


By

Dr Kathryn E. Merrick

Senior Lecturer in Information Systems and Computer Science

University of New South Wales, Australian Defence Force Academy, Canberra, ACT, Australia

Abstract:
Learning skills online from experiences is attractive for robots because it permits a robot to develop its skills autonomously. However, the onus lies with the system designer to specify which skills the robot should learn. Experience-based goal generation algorithms permit a robot to decide autonomously which skills to learn. However, such algorithms have not been widely explored in robotics. This talk presents an approach to experience-based generation of achievement and maintenance goals for a mobile robot and discusses how these goals can be used to motivate learning. The talk will cover the design of an experience-based goal generation algorithm and experimental analysis of the algorithm for generating different types of goals. A selection of reinforcement learning algorithms for learning skills to solve self-motivated goals will also be discussed.

Short biography:
Dr Kathryn E. Merrick, received Bachelor degree in Computer Science and Technology (Advanced, Honours I, University Medal) at University of Sydney, NSW, Australia, 2002; PhD in Computer Science from National ICT Australia and University of Sydney, NSW, Australia, 2007. 
Dr Merrick is a Senior Lecturer in Information Systems and Computer Science at the University of New South Wales, Australian Defence Force Academy, Canberra, ACT, Australia. Her research interests lie in the broad areas of artificial intelligence and machine learning, with applications in virtual characters, developmental robotics and intelligent environments. Her research is principally concerned with the development of algorithms and metrics for self-motivated learning agents. She is co-author of the book Motivated Reinforcement Learning: Curious Characters for Multiuser Games (Berlin: Springer-Verlag, 2009) and over forty refereed conference and journal papers.

Dr Kathryn E. Merrick Normal 0 false false false EN-GB X-NONE AR-SA

Braitenberg vehicles: From Theory to Implementation

Date: Thursday 21/11/2013
Venue: MS020
Time: 1.00 pm
Speaker: Dr. Inaki Rano
Affiliation: School of Computing and Intelligent Systems
University of Ulster, Magee


Braitenberg vehicles: From Theory to Implementation


By

Dr. Inaki Rano

Lecturer in Cognitive Robotics

School of Computing and Intelligent Systems
University of Ulster, Magee

 

Abstract:
Braitenberg vehicles are well known models of animal behaviour used assteering mechanisms in mobile robotics and Artificial Life. Because of their simplicity, they are mainly used for teaching robotics, whilst the lack of a quantitative theory makes troublesome their use for research purposes. This talk will present ongoing efforts to quantitatively model, understand and use Braitenberg vehicles. Starting from the qualitative model the talk will present theoretical results and an implementation of one of the vehicles which can be used for coverage tasks like cleaning, de-mining or surveillance.

Short biography:
Inaki Rano holds a MSc in Physics ('97) and a PhD in Computer
Sciences ('04) from the University of the Basque Country. From 1997 to
2004 and from 2005 to 2010 he was, respectively, a member of the
Robotics and Autonomous Systems Group of the University of the Basque
Country, and the Robotics, Perception and Real Time Group of the
University of Zaragoza. In 2011 he joined the Institute for Neural Computation
of the Ruhr-Universitat Bochum. Since 2013 he is Lecturer on Cognitive
Robotics in the School of Computing and Intelligent Systems at the University
of Ulster. His main research interest focus on bio-robotics and control theory.

Development of the RUBICON (Robotic UBIquitous COgnitive Network) cognitive architecture

Date: Thursday 25/4/2013
Venue: MS020
Time: 13.00 pm
Speaker: Dr. Anjan Kumar Ray
Affiliation: ISRC, University of Ulster, Magee


Development of the RUBICON (Robotic UBIquitous COgnitive Network) cognitive architecture
By

Dr. Anjan Kumar Ray

ISRC, University of Ulster, Magee

Abstract:


The RUBICON (Robotic UBIquitous COgnitive Network) is an EU FP7 funded project. It aims to create a self-sustaining and learning robotic ecology to support applications such as smart home environment. Information extraction from sensors, robotic interactions with the environment and building knowledge over this information are the key aspects of this project. There are four technical layers named learning, control, cognitive and communication layers to achieve the purpose of the ecology. Our focus is on the development of the cognitive architecture for the system. We explore the potential of a self-organizing fuzzy neural network (SOFNN) as a core component of a cognitive system for a smart home environment. We develop a cognitive reasoning module that has the ability to adapt its neuronal structure through adding and pruning of neurons according to the incoming data. The network is trained with realistic synthesized data to show its adaptation capability and is tested with unseen data to validate its cognitive capabilities. This initial implementation of the cognitive system demonstrates the potential of the architecture and will serve as a very important test-bed for future work.




Short biography:

Anjan Kumar Ray is a Research Associate in the ISRC. He is currently working in the RUBICON project. He received his PhD from Indian Institute of Technology (IIT), Kanpur and ME in Control Systems from Bengal Engineering and Science University, Shibpur. His research interests are in the areas of robotics, machine learning and control systems.

Evolving Spiking Neural Networks: Methods, Systems and Applications for Spatio- and Spectro-Temporal Pattern Recognition

Date: Thursday 31/10/2013
Venue: MS020
Time: 12.00 pm
Speaker: Prof. Nikola Kasabov
Affiliation: Director, Knowledge Engineering and Discovery Research Institute (KEDRI)
Auckland University of Technology



Evolving Spiking Neural Networks: Methods, Systems and Applications for Spatio- and Spectro-Temporal Pattern Recognition
By

 Prof. Nikola Kasabov

Director, Knowledge Engineering and Discovery Research Institute (KEDRI)
Auckland University of Technology




Abstract:


Spatio- and spectro-temporal data (SSTD) are the most common data in many domain areas, including: signal processing; bioinformatics; neuroinformatics; ecology; environment; medicine; economics, etc., and still there are no efficient methods to model such data and to discover complex spatio-temporal patterns from it. The talk introduces new methods for modeling and pattern recognition of SSTD based on evolving spiking neural networks (eSNN). eSNN develop their structure and functionality from streams of data in an on-line learning mode [1]. They consist of: input encoding module; 3D SNN reservoir structure; eSNN classifier. They can include also gene regulatory networks as [2]. Different eSNN models are presented, such as: probabilistic neuronal model [3], the dynamic evolving SNN (deSNN) [4], SPAN [5], reservoir SNN NeuCube [6]; quantum inspired SNN [7] and others.
Applications across domain areas are demonstrated, including: moving object recognition [4]; integrated audio-visual pattern recognition [7]; EEG data modeling [8]; design of artificial cognitive and emotional systems [10]. Challenging open problems and future directions are presented [11,12] .
Software systems for building SNN are demonstrated introduced the KEDRI Python EvoSpike simulator and the KEDRI MATLAB NeuCube simulator.

Short biography:

Professor Nikola Kasabov is Fellow of the IEEE, Fellow of the Royal Society of New Zealand and Distinguished Visiting Fellow of the Royal Academy of Engineering. He is the Director of the Knowledge Engineering and Discovery Research Institute (KEDRI), Auckland. He holds a Chair of Knowledge Engineering at the School of Computing and Mathematical Sciences at Auckland University of Technology. Kasabov is a Past President and Governors Board member of the International Neural Network Society (INNS) and also of the Asia Pacific Neural Network Assembly (APNNA). He is a member of several technical committees of IEEE Computational Intelligence Society and a Distinguished Lecturer of the IEEE CIS. He is a Co-Editor-in-Chief of the Springer journal Evolving Systems and has served as Associate Editor of Neural Networks, IEEE TrNN, IEEE TrFS, Information Science, J. Theoretical and Computational Nanosciences, Applied Soft Computing and other journals. Kasabov holds MSc and PhD from the TU Sofia, Bulgaria. His main research interests are in the areas of neural networks, intelligent information systems, soft computing, bioinformatics, neuroinformatics. He has published more than 510 publications that include 15 books, 160 journal papers, 80 book chapters, 28 patents and numerous conference papers. He has extensive academic experience at various academic and research organisations in Europe and Asia, including: TU Sofia, University of Essex, University of Otago, Guest professor at the Shanghai Jiao Tong University, Guest Professor at ETH/University of Zurich. Prof. Kasabov has received the APNNA ‘Outstanding Achievements Award’, the INNS Gabor Award for ‘Outstanding contributions to engineering applications of neural networks’, the EU Marie Curie Fellowship, the Bayer Science Innovation Award, the APNNA Excellent Service Award, the RSNZ Science and Technology Medal, and others. He has supervised to completion 35 PhD students. More information of Prof. Kasabov can be found on the KEDRI web site: http://www.kedri.aut.ac.nz.

Dealing with non-stationary EEG signals in brain-computer interface


Date: Friday 1/3/2013
Venue: MS020
Time: 12.00 pm
Speaker: Ms. Mahnaz Arvaneh
Affiliation: Institute for Infocomm Research (I2R), Singapore


Dealing with non-stationary EEG signals in brain-computer interface

Ms. Mahnaz Arvaneh

Institute for Infocomm Research (I2R), Singapore

 

Abstract:

A major challenge in EEG-based brain-computer interfaces (BCIs) is the inherent non-stationarities in the EEG data. Variations of the signal properties from intra and inter sessions often lead to deteriorated BCI performances. To address this issue, we proposed two novel algorithms.
The first algorithm focuses on intra-session non-stationarities by extracting features that are robust and invariant. This was achieved by a novel spatial filtering algorithm, called KLCSP. The proposed KLCSP algorithm simultaneously maximizes the discrimination between the class means, and minimizes the within-class dissimilarities measured by a loss function based on the Kullback-Leibler (KL) divergence. The proposed KLCSP algorithm is completely data-driven, and only uses the train data to make the features robust and invariant against non-stationarities and variations.
The second algorithm focuses on inter-session non-stationarities by proposing a novel data space adaptation technique, called EEG data space adaptation (EEG-DSA). The EEG-DSA algorithm linearly transforms the EEG data from the evaluation session, such that the distribution difference to the training session is minimized. Using the KL divergence criterion, two versions of the EEG-DSA algorithm are proposed: the supervised version when labelled data are available in the evaluation session, and the unsupervised version when labelled data are not available. By adapting the EEG data space directly, the proposed EEG-DSA algorithm is not limited to any specific BCI models. In contrast, it can be applied with other techniques which adapt the feature or classifier spaces. In addition, the proposed algorithm can be potentially applied on other spaces, such as ECoG, MEG etc.


Short biography:


Mahnaz Arvaneh received the B.Sc. degree in electrical engineering from K. N. Toosi University of Technology, Tehran, Iran, and the M.Sc. degree in control engineering from Ferdowsi University of Mashhad, Iran, in 2005 and 2007, respectively. She is currently pursuing the Ph.D. degree with Singapore Nanyang Technological University. Since April 2009, she is an attached student in Institute for Infocomm research, Agency for Science, Technology and Research, Singapore. In 2012, she was a visiting student with the Electrical Engineering Department, National University of Ireland, Maynooth, Ireland. Her current research interests include brain–computer interfaces, signal processing, machine learning, and pattern recognition.