ISG@Macquarie

        Abhaya Nayak
            Intelligent Systems Group

Home

Contact

Bio-sketch

Teaching

Research

Publications

Catch-all

 
 

Om Shantih! Shantih! Shantih Om!  

Festschrift for Prof Norman Foo

Belief Revision

Sydney Chatashali -- Oriya Language and Culture Down Under

NEWS LINKS

Samachar -- Indian News

Sydney Morning Herald

Washington Post

 

 

 

Research

 

PhD Scholarships available

About my research

Some Research Projects

Recent Research Grants

PhD SCHOLARSHIPS AVAILABLE

Nothing at the moment, but please check later.

ABOUT MY RESEARCH

I study the doxastic (that is Intellectualese for belief-related) aspects of intelligent systems. In particular, I study the dynamics of belief systems.

If you search the web for the target string "belief system", most of the sites you will hit would deal with belief systems associated with different religions. The belief systems I am interested in have very little or nothing to do with religions. A belief system, for me, is a formal system for carrying out representation of the beliefs of an agent, and reason from/about them.

A very simplified way of looking at beliefs is the following: given its goals, an agent reasons from its beliefs about its current state and its environment, to what actions it should perform. Thus, given that I want to drive to the air port, and that I believe the most direct route would be very crowded, I might reason and take the action of driving along a back-route. Thus beliefs are intimately connected with actions.

Here is a very rudimentary account of intelligent system. The system can observe its environment. The observations are used to update its beliefs about the environment. Its changed beliefs might affect the actions it wants to take to reach its goals. AND, the actions it performs modifies its environment, whereby new observations mandate modification in its beliefs. Almost any intelligent system can model this scenario. For instance, the thermostat in your refrigerator can be an intelligent system. It makes an observation (temperature); updates its beliefs accordingly (Is it cold enough?), based on its beliefs and the goal (maintain temperature at four degrees), it might perform some action (eg., shut off the refrigerator) that impacts the environment which might lead to different observations.

There are many interesting aspects to this simple scenario.

  1. You might assume that the system in question never makes a faulty observation. In other words, if two instances of the same test result in different observations, then it must have been because the environment has undergone some change in the mean time.
  2. You might make an assumption on the other extreme. That if the later observation differs from the former, it is not because the environment has changed, but because one (or both) of the observations are inaccurate. In other words, the environment is static, and the observations are all potentially faulty.
  3. You can also relax both these assumptions -- you allow the domain to be dynamic and the observations inaccurate.
  4. To enhance the performance of the refrigerator, you might want the thermostat to receive input from five sensors instead of one, the sensors being positioned at different parts of the refrigerator.
  5. There is no reason why all the sensors in the refrigerator will measure temperature alone! You might want some of the sensors to provide humidity information, and the action of the refrigerator depend on its beliefs both about the temperature and the humidity.
As we change the domain, the issues involved get more complicated. But there is room enough for a lot of good research even under idealistic assumptions. Here are some of the related research topics that I am interested in
  • Reasoning about action and change,
  • Belief revision and belief update
  • Reasoning under uncertainty and/or inconsistency
  • Preference representation and Belief Merging
  • Planning under incomplete knowledge
  • Intelligent Information Assimilation
  • Information aggregation and social choice
  • Coalition formation among information agents

SOME RESEARCH PROJECTS

Intelligent Information Assimilation

This project is supported by an ARC Discovery Grant (2003-05). Dr Maria del Pilar Pozos Parra, a research fellow in the Intelligent Systems Group, is supported by this grant.

Brief Description. Making intelligent and informative choice usually involves assimilating information from multiple sources. Not all sources are however mission critical -- information from some of the sources may be more important than others. Besides, not all sources can be trusted to the same degree. This makes the task of assimilating such information very difficult. Recent research on belief merging provides a formal framework designed to facilitate this task. In this project we will

  • examine existing and novel belief merging strategies
  • study their impact on allied disciplines
  • extend the framework for suitable applications
  • design, implement and test fielded applications of this framework.
The outcome of this research will provide the basis for many industrial applications such as agent negotiation

Intelligent Applications through the Semantic Web

This is not really a "project". This is an endeavour to build a network of researchers in Australia and overseas whose research interest relate to this broad area of research. This endeavour has received $20,000 from the Australian Research Council as seed funding money to prepare the proposal. It has received further support of $20,000 from different units of the Macquarie University. Should you be interested to know more about it, please let me know.

Brief Description. The primary aim of the proposed bid is to build a network of Australian researchers and their international peers for conducting research into the fundamental as well as applied aspects of the Semantic Web. By incorporating meaning of web-content in a form that can be accessed and processed by intelligent software agents, the Semantic Web will allow computers and humans to work in cooperation. This research will address the needs of both the Australian government and industry that provide and make smart use of information available on the Web. It will ensure Australian preparedness for the next-generation web technology.

RECENT RESEARCH GRANTS

2008

  1. Nayak, A. and P. Busch, "Formal Foundations for Tacit Knowledge: Elicitation, Representation and Reasoning", Macquarie University Research Development Grant (2008), $24,710.
  2. Zhang, Y., A. Nayak, K. Wang and F. Lin, "Foundations of Nonmonotonic Logic Programming for Complex Knowledge Systems", ARC Discovery Grant (2008), $76,000.
2007
  1. Nayak, A. and P. Busch, "Formal Foundations for Tacit Knowledge: Elicitation, Representation and Reasoning", Macquarie University Research Development Grant (2007), $23,429.
  2. Zhang, Y., A. Nayak, K. Wang and F. Lin, "Foundations of Nonmonotonic Logic Programming for Complex Knowledge Systems", ARC Discovery Grant (2007), $74,000.
  3. Zhang, Y., M. Orgun, A. Nayak, Y. Mu and F. Bao, "Knowledge Based Model Updating for the Correctness of Security Protocols", ARC Discovery Grant (2007), $82,000.
2006
  1. Zhang, Y., A. Nayak, K. Wang and F. Lin, "Foundations of Nonmonotonic Logic Programming for Complex Knowledge Systems", ARC Discovery Grant (2006), $84,000.
  2. Zhang, Y., M. Orgun, A. Nayak, Y. Mu and F. Bao, "Knowledge Based Model Updating for the Correctness of Security Protocols", ARC Discovery Grant (2006), $80,000.
2005
  1. Zhang, Y., M. Orgun, A. Nayak, Y. Mu and F. Bao, "Knowledge Based Model Updating for the Correctness of Security Protocols", ARC Discovery Grant (2005), $122,000.
  2. Nayak, A., N. Foo, A. Ghose and M. Pagnucco, "Intelligent Information Assimilation", ARC Discovery Grant (2005), $55,000.
  3. Pagnucco, M., C. Sammut, A. sattar and A. Nayak, "Real-time high-level cognitive robotics controllers", ARC Discovery Grant (2005), $70,000. (Relinquished.)
2004
  1. Seed-funding for ARC Research Network on Intelligent Applications through the Semantic Web. (2004). $40,000 ($20,000 from the ARC plus $20,000 from several parts of the Macquarie University).
  2. Nayak, A., N. Foo, A. Ghose and M. Pagnucco, "Intelligent Information Assimilation", ARC Discovery Grant (2004), $60,000.
  3. Pagnucco, M., C. Sammut, A. sattar and A. Nayak, "Real-time high-level cognitive robotics controllers", ARC Discovery Grant (2004), $80,000. (Relinquished.)
2003
  1. Nayak, A., N. Foo, A. Ghose and M. Pagnucco, "Intelligent Information Assimilation", ARC Discovery Grant (2003), $69,000.
  2. Pagnucco, M., C. Sammut, A. sattar and A. Nayak, "Real-time high-level cognitive robotics controllers", ARC Discovery Grant (2003), $90,000. (Relinquished.)
  3. Nayak, A., et al Research Infrastructure Block Grant (2003) under DEST, $30,000.
2002 and Before
  1. Nayak, A. and Orgun, M. Macquarie University, MURG (2002) "Information assimilation and smart choice", $6,877.
  2. Nayak A. and Pagnucco M. Macquarie University, Research Development Scheme Grant (previously ARC Small-Grant) 2001. "Discovering Causal Laws Through Under-specified Actions" $12,100.
  3. Nayak, A. Macquarie University, MURG (2001) "Intelligent Information Assimilation and Choice Making Based on Belief Merging Technique" $6,800.
  4. Pisan, Y and Nayak A. Macquarie University, MURG (2001) "Using Multi-Agent Environments Based on Real World Models for Training" $6,500.


¨ 2003 Macquarie University