Transcript IntelligentAgents

Current research in
Intelligence Agents
Victor Govindaswamy
Contents
• Introduction
• Example uses of Agents in
– World Wide Web
– Water systems
– Smart Cars
– Virtual reality-based training
– Medical image processing systems
• Conclusion
Introduction
• Roles of intelligence agents have expanded
since the 1990s…
• Variety of roles
• These roles are demonstrated in recent projects
such as
–
–
–
–
–
–
Smart Cars,
World Wide Web,
Medical image processing,
Monitoring and analyzing processes,
Virtual reality-based training systems
Water systems.
Example use of Agents in
World Wide Web
Sapp, D.; Shang, Y., Intelligent Web representatives, Proceedings 11th
International Conference on Tools with Artificial Intelligence 1999.
• Introduced a design and implementation of
an intelligent agent that is based on
existing Internet and Web technologies
• 3 main design parts to fulfill the need
– for a natural language parser,
– to represent knowledge and
– for effective communication between the user
and the computer.
Example use of Agents in
World Wide Web
Sapp, D.; Shang, Y., Intelligent Web representatives, Proceedings 11th
International Conference on Tools with Artificial Intelligence 1999.
•
Splits a sentence into its subject, verb and predicate using a parser
– 3 main parts for the parser’s algorithms:
• Splits apart the verb by using a database of known verbs
• Search the unknown subject and predicate for keywords
• Return the contents of the sentence in a usable format
•
The agent narrows down the user’s question to a few key terms
•
Pattern matched to find the best answer
•
Uses Extensible markup language (XML)
– a language was developed based on XML for searching the database and a
separate parser to sift through the content and locate the answer.
•
As for communication between user and communication, Microsoft Agent is used
– voice recognition and other advanced features.
Example use of Agents in
Smart Cars
Bourbakis, N.; Findler, M., Smart cars as autonomous intelligent agents,
Proceedings 13th IEEE International Conference on Tools with Artificial
Intelligence, 2001.
• Propose an idea to use intelligent agents in a
smart car to gather information related to its
surrounding environment
• Agents are being used to
– determine its position and distance relative to other
moving and stationary objects
– detect motion and track behavioral patterns of other
objects
– exchange information via internet with the other
moving objects in order to cooperate and negotiate a
safe environment during the journey
Example use of Agents in
water sharing
Le Bars, M.; Attonaty, J.M., A multi-agent system to the common management
of a renewable resource: application to water sharing, Proceedings 13th IEEE
International Conference on Tools with Artificial Intelligence, 2001
• Propose that intelligent agents can be used to solve the water
sharing problem among farmers in France
• Current approaches such as linear programming or game theory are
based on assumptions that
– the people involved are completely rational and include only a limited
number of people..
– Not dynamic
• Building an Agent-Based Modeling (ABM) with a Multi-Agent
approach that enables negotiations among a variety and an everincreasing number of players by taking into consideration
– water allocation rules,
– the players’ changing attitude and behavior.
Example use of Agents in
Virtual Reality-based Training Systems
Fuhua Lin; Chuan-Jun Su; Tseng, M.M., An agent-based approach to developing
intelligent virtual reality-based training systems, Proceedings 11th International
Conference on Tools with Artificial Intelligence, 1999.
• Introduce an agent-based approach to modeling intelligent Virtual
Reality-based Training Systems (VRTS)
– to help companies train their employees.
• The characteristics of this approach includes
– An agent-based architecture,
– Using Petri Nets modeling is also used to realize communication,
coordination and task control among agents, and task plan designing,
pedagogical knowledge modeling, and behavior modeling of virtual
objects.
– Several agents are being used for training-task planning, simulation,
performance evaluation, instruction and interfacing.
Example use of Agents in
interpreting medical images
Popescu, M.; Yi Shang, An agent-based approach for interpreting medical images,
Proceedings 11th International Conference on Tools with Artificial Intelligence, 1999 .
• Use two types of agents in their agentbased approach for interpreting medical
images
– patient representative and
• receives an image of the patient from a web-based interface and
send it to a radiologist agent
– radiologist
• interprets the image using different image analysis algorithms
• has a knowledge of anatomy that applies to all imaging modalities
and know-how of the specific physics of every modality in order to
interpret a given image
Conclusion
• All these projects are trying to make the
agents more intelligent than ever.
• Although the agents’ objectives might be
different and specific to a task, humankind
are inching their way to the ultimate goal
of making agents as intelligent or better
yet, more intelligent than humans.