| Intelligent Agents | ||
|
Intelligent
Agents Project at IBM T.J. Watson Research
Application
of Intelligent Agents An Outline of Intelligent Agents An Online List of
all Agents Agents
101 "Is
there an Agent in Your Future" Conference
on Autononous Agents "Where
do Intelligent Agents come from?" Every
link you might ever need "Intelligent
Agents: A Primer"
|
An agent is a computer system
placed in an environment that is capable and free of a dominant
controller. It acts without direct intervention of humans.
Autonomous agents are software entities that are capable of independent
action in dynamic, unpredictable environments. Agents are also one of the
most important areas of research and development in computer science
today. Agents are currently being applied in domains as diverse as
computer games and interactive cinema, information retrieval and
filtering, user interface design, and industrial process control.
Agent-based software has been called the "next significant breakthrough in software development." Agents are the focus of strong interests on the part of computer science and artificial intelligence. Agents are used in such things as ranges between e-mail filters to air traffic control. Other examples are: * Music and movie recommendation system * News filters * Literary filters * Webpage Pop-UP stoppers * Search engines that provide automated "BOTS" A good agent has the following characteristics: (1) you can communicate with it properly, (2) the agent can act as well as suggest topics/answers, (3) an agent can act without supervision and (4) it can use experience to help you. In other words "It must be communicative: able to understand your goals, preferences and constraints. It must be capable: able to take options rather than simply provide advice. It must be autonomous; able to act without the user being in control the whole time. And it should be adaptive; able to learn from experience about both its tasks and about its users preferences." Critics argue that a third-party might be able to take control of an agent and provide information that would merely benefit them, not the consumer. Other people believe that a human supervisor should be implemented to monitor some of the actions that occur. While the software may provide as that supervisor for the system it In order for the system to be adaptive behavior can be assessed in a number of ways. The simplest way is to group users based on some set of features, and then to assume similarity between them. This can work fairly easily. A new user can fill out a questionnaire that allows the system (using a statistical clustering algorithm) to figure out to which cluster of other users this user belongs. Preferences associated with that group are then assumed to work for this user. How They Work: Any of several technologies can design intelligent agents. All of them use some combination of statistical operations, artificial intelligence, machine learning, inference, neural networks, and information technologies. Agent systems are not plug and play. They need to be trained or taught. Most require examples of right answers or rules for appropriate behavior. Typically, an agent system is implemented in several stages. First, one develops rules or training data. Then, one either trains the agents by giving them rules such as, “If Maes writes an article, then get the whole article and notify me by flashing the new articles icon when it arrives,” or by giving them a large set of examples with the “right” answers included. Once the agent system performs satisfactorily on the training data, it is ready to work on test data to make sure that it can extend what it has learned to unknown materials. A last step, but a continuing one, at least in theory, is to evaluate performance at several intervals. Agents should learn over time, and their performance should improve as they adapt to the user’s needs, as well as to the kinds of information they navigate. The Questions that are involved in this process: Trust
|