Paper Review: Learning to Manipulate and Categorize in Human and Artificial Agents

#Title#

Learning to Manipulate and Categorize in Human and Artificial Agents

#Authors#

Giuseppe Morlino, Claudia Gianelli, Anna M. Borghi, Stefano Nolfia

#Venue#

Cognitive Science (2014) 1–26

#DOI#

DOI: 10.1111/cogs.12130

#Abstract#

This study investigates the acquisition of integrated object manipulation and categorization abilities through a series of experiments in which human adults and artificial agents were asked to learn to manipulate two-dimensional objects that varied in shape, color, weight, and color intensity. The analysis of the obtained results and the comparison of the behavior displayed by human and artificial agents allowed us to identify the key role played by features affecting the agent/environment  interaction, the relation between category and action development, and the role of cognitive biases originating from previous knowledge.

#Comments#

The paper looks at issues in the effect of action on categorisation.  They present that categorisation is grounded in in the sensorimotor system, according to present experiments and theory.  And again suggest the central role of action in cognition.

They also look at the issues around how categories enable the flexible usage of objects, and how the grasping of objects changes according to the tasks needed, as per the classic idea of affordances by Gibson (1979).

Important quote: "Affordances are proposed to be the product of the conjunction, in the brain, of repeated visuomotor experiences." Probably a no-brainer to the design community, but important to me, as I need to see this generalise to virtual worlds.  It should be noted that the systems used in this experiment were synthetic, so the effects should generalise to a virtual world, as it is simply shapes and colours with physical properties.  However, there is a history of visual search research with simple shapes not generalising to real images.  This must be considered in any assumptions of efficacy in virtual world simulations.

The experiments involved the manipulation of 2D objects on the screen with a mouse pointer in placing and shaking tasks.   The weight of the objects is aligned with categories and some of the categories are also based on colour, blinking and shape.  The humans (20) were compared to neural network agents.

"The results indicated the discriminative features affecting the agent environment interaction such as weight facilitate the acquisition of the required categorisation abilities with respect to alternative features that are equally informative but that do not affect the outcome of the agent actions."  This leads them to the conclusion that the categorisation for both humans and agents, not withstanding any other factors, is affected by the embodiment of the activity; weight required interaction, not just observation.

The results showed support for a model whereby the interaction with light vs heavy objects produces categories far more effectively than other factors.  Embodied action thus has a great affect on categorisation, whether it affects every category is still uncertain, as the other visual effects (from grounded cognitive affects) still caused categories to form, just not as soon in the training.

They consider this to contribute to a STRONG position of embodiment being central to the creation of categories, and not just being a more peripheral contributor.

They also note a shape effect with humans, ie. they used a curvilinear path with circles, and a rectilinear path with square.  Thus previous memories of the objects influenced their actions and thus the categories.

They also note that the categories are from an interaction of the agent with the environment, and not so from top-down or bottom-up processes exclusively, not overgeneralised or fine granularity categories, but as a dynamic process between agent and environment.

While this is categorisation, and not a memory task, one still has to wonder, for my work, if the memory of a process will be much more enhanced by embodied interactions, and not just visual interactions alone.  One could hypothesise that if the category is more strongly created with embodied action, then the memory of that category (if it maps to say activity specifications) then should be stronger on acting it out.  So an Occulus and Kinect space should measurably work better in process memory tasks than a pure visual space; with both working better than a simple interview.

Something to think about I guess.

#ImportantRefs#

Paper: Neural network-based detection of virtual environment anomalies

Just had a paper Neural network-based detection of virtual environment anomalies accepted to Neural Computing and Applications, with Alfredo Nantes and Frederic Maire.  The paper is here at QUT ePrints, and here at Springer.

Abstract The increasingly widespread use of large scale 3D virtual environments has translated into an increasing effort required from designers, developers and testers. While considerable research has been conducted in assisting the design of virtual world content and mechanics, to date, only limited contributions have been made regarding the automatic testing of the underpinning graphics software and hardware. In this work, two novel neural network-based approaches are presented to predict the correct visualization of 3D content. Multi-Layer Perceptrons and Self Organizing Maps are trained to learn the normal geometric and color appearance of the objects from validated frames, and used to detect novel or anomalous renderings in new images. Our approach is general, for the appearance of the object is learned rather than explicitly represented. Experiments were conducted on a game engine to determine the applicability and effectiveness of our algorithms. The results show that the neural network technology can be effectively used to address the problem of automatic and reliable visual testing of 3D virtual environments.

Well done Alfredo!

Ross

CFP: PRIMA 2012 Call for Papers

CALL FOR PAPERS

PRIMA 2012

The 15th International Conference on Principles and Practice of Multi-Agent Systems

Kuching, Sarawak, Malaysia

September 3-7, 2012

http://www.prima2012.org

Co-located with PRICAI 2012, DC 2012 and PKAW 2012

Important Dates

Workshop proposals

December 15, 2011

Papers/Tutorial proposals

March 30, 2012

Author notification

May 28, 2012

Camera-ready papers

June 15, 2012

Workshops and Tutorials

September 3-4, 2012

Conference

September 5-7, 2012

Agent computing is an exciting, transformational approach to developing computer systems that can rapidly and reliably solve real-world problems that usually demand human knowledge and expertise. The value, power and flexibility of agent and multi-agent systems has been demonstrated in application areas such as logistics, manufacturing, simulation, robotics, decision support, entertainment, and especially in online market environments. As one of the largest and fastest growing research fields of Computer Science, agent research today includes a wealth of topics. The PRIMA 2012 Program Committee invites submissions of original, unpublished, theoretical and applied work on any such topic, and encourages reports on the development of prototype and deployed agent systems, and of experiments that demonstrate novel agent system capabilities.

Submission. PRIMA 2012 proceedings will be published by Springer as a volume in the LNAI series and proceedings will be available at the conference. Submitted papers should be 12–15 pages in Springer LNCS format and must be in a form suitable for "double-blind" review. Each submission will be subject to peer review in two rounds coordinated by an international Senior Program Committee, and authors will be able to provide a short "rebuttal" of the reviews before final decisions are made. A broad range of agent topics are of interest, but all papers should clearly identify how their scientific or technical contributions advance the state-of-the-art of agent computing practice or have a strong potential to do so Submitted papers should not be under review or submitted for publication elsewhere during the review period.

Springer LNCS Author Instructions: http://www.springer.de/comp/lncs/authors.html

Submission: http://www.easychair.org/conferences/?conf=prima2012

Enquiries: prima2012-pc-chairs@cse.unsw.edu.au

Organization

General Chairs

Sandip Sen (University of Tulsa, USA)

Toshiharu Suguwara (Waseda University, Japan)

Local Arrangements Chairs

Dickson Lukose (MIMOS Berhad, Malaysia)

Cheah Wai Shiang (Universiti Malaysia Sarawak, Malaysia)

Sponsorship Chairs

Longbing Cao (University of Technology, Sydney, Australia)

Matthias Klusch (DFKI Saarbruecken, Germany)

Sarvapali Ramchurn (University of Southampton, UK)

Jie Zhang (Nanyang Technological University, Singapore)

Tutorial Chair

Edith Elkind (Nanyang Technological University, Singapore)

Workshop Chairs

Sherief Abdallah (British University in Dubai, UAE and

University of Edinburgh, UK)

Hiromitsu Hattori (Kyoto University, Japan)

Program Chairs

Iyad Rahwan (Masdar Institute, UAE and MIT, USA)

Wayne Wobcke (University of New South Wales, Australia)

Senior Program Commitee

Stephen Cranefield (University of Otago, New Zealand)

Frank Dignum (Utrecht University, The Netherlands)

Guido Governatori (NICTA, Australia)

Katsutoshi Hirayama (Kobe University, Japan)

Kate Larson (University of Waterloo, Canada)

Rey-Long Liu (Tzu Chi University, Taiwan)

Alessio Lomuscio (Imperial College London, UK)

Andrea Omicini (University of Bologna, Italy)

Jeremy Pitt (Imperial College London, UK)

David Pynadath (University of Southern California, USA)

Alex Rogers (University of Southampton, UK)

Paolo Torroni (University of Bologna, Italy)

Publicity Chairs

Jacob Crandall (Masdar Institute, UAE and MIT, USA)

Koen Hindriks (Delft University of Technology, The Netherlands)

Topics

Foundations of Agents and Multi-Agent Systems

Logics of Agency

Logics of Multi-Agent Systems

Normative Systems

Computational Game Theory

Uncertainty in Agent Systems

Agent and Multi-Agent Learning

Agent and Multi-Agent System Architectures

Agent Programming Languages and Platforms

Multi-Agent System Languages and Platforms

BDI Architectures and Extensions

Normative Multi-Agent Systems

Agent-Oriented Software Engineering

AOSE Methodologies

Tools for Agent and Multi-Agent System Development

Formal Specification and Verification

Deployed System Case Studies

Agent-Based Modelling and Simulation

Simulation Languages and Platforms

Artificial Societies

Virtual Environments

Workflow Simulation

Emergent Behaviour

Modelling System Dynamics

Application Case Studies

Collaboration/Coordination/Communication

Agent Communication Languages and Protocols

Distributed Problem Solving

Teamwork Models

Coalition Formation

Argumentation, Negotiation, Bargaining

Auctions and Mechanism Design

Trust and Reputation

Computational Voting Theory

Hybrid Technologies

Agents in Planning

Agent-Based Scheduling

Agent-Based Optimization

Distributed Constraint Satisfaction

Agents and Data Mining

Semantic Web Agents

Agents and Grid Computing

Agents and Service Oriented Computing

Agents and Pervasive Computing

Robotics and Multi-Robot Systems

Application Domains

Healthcare

Transport/Logistics

Emergency/Disaster Management

Energy/Utility Management

Sustainability/Resource Management

Games/Entertainment

eBusiness/eCommerce/eGovernment

eResearch/eLearning

Security/Surveillance

Smart Cities

Applications

Adaptive Personal Assistants

Embodied Conversational Agents

Virtual Characters

Multi-Modal User Interfaces

Autonomous Systems

Mobile Agents

Human-Robot Interaction

Social Recommender Systems

Paper: Human resource behaviour simulation in business processes

Rune Rasmussen and I have just had a new ERA A journal paper accepted with Future Generation Computer Systems (FGCS). QUT EPrints Link

Title: Human resource behaviour simulation in business processes

Abstract: The structure and dynamics of a modern business environment are very hard to model using traditional methods. Such complexity raises challenges to effective business analysis and improvement. The importance of applying business process simulation to analyze and improve business activities has been widely recognized. However, one remaining challenge is the development of approaches to human resource behavior simulation. To address this problem, we describe a novel simulation approach where intelligent agents are used to simulate human resources by performing allocated work from a workflow management system. The behavior of the intelligent agents is driven a by state transition mechanism called a Hierarchical Task Network (HTN). We demonstrate and validate our simulator via a medical treatment process case study. Analysis of the simulation results shows that the behavior driven by the HTN is consistent with design of the workflow model. We believe these preliminary results support the development of more sophisticated agent-based human resource simulation systems.

Ross

Paper: Human resource behaviour simulation in business processes

Our PhD student Hanwen Guo has been busily working away on simulating human agent behaviours with regards to workflow systems such as YAWL. He has now developed and tested an HTN-based set of agents for testing resource models in virtual worlds. Preliminary results are available in an ISD (Era A) conference paper here, applied to a health care workflow scenario.

Well done Hanwen!

Ross

Abstract

The structure and dynamics of a modern business environment are very hard to model using traditional methods. Such complexity raises challenges to effective business analysis and improvement. The importance of applying business process simulation to ana- lyze and improve business activities has been widely recognized. However, one remaining challenge is the development of approaches to human resource behavior simulation. To ad- dress this problem, we describe a novel simulation approach where intelligent agents are used to simulate human resources by performing allocated work from a workflow manage- ment system. The behavior of the intelligent agents is driven a by state transition mechan- ism called a Hierarchical Task Network (HTN). We demonstrate and validate our simulator via a medical treatment process case study. Analysis of the simulation results shows that the behavior driven by the HTN is consistent with design of the workflow model. We be- lieve these preliminary results support the development of more sophisticated agent-based human resource simulation systems.