small lecturenumber - hepage : Outlinesmall lecturenumber - hepage : Distributed AIsmall lecturenumber - hepage : Qualities of an agentsmall lecturenumber - hepage : Autonomysmall lecturenumber - hepage : Agent-oriented Programmingsmall lecturenumber - hepage : Usefulness of the agent metaphorsmall lecturenumber - hepage : Agents and Servicessmall lecturenumber - hepage : Why solve DAI problems?small lecturenumber - hepage : Some examplessmall lecturenumber - hepage : Recommender Systemssmall lecturenumber - hepage : Content-based Filteringsmall lecturenumber - hepage : Content-based Filteringsmall lecturenumber - hepage : Collaborative Filteringsmall lecturenumber - hepage : Collaborative Filteringsmall lecturenumber - hepage : Collaborative Filteringsmall lecturenumber - hepage : Coordinationsmall lecturenumber - hepage : Mechanisms for Coordinationsmall lecturenumber - hepage : Mechanisms for Coordinationsmall lecturenumber - hepage : Mechanisms for Coordinationsmall lecturenumber - hepage : Mechanisms for Coordinationsmall lecturenumber - hepage : Cooperative environmentssmall lecturenumber - hepage : Contract Netsmall lecturenumber - hepage : Contract Netsmall lecturenumber - hepage : Example: distributed sensor managementsmall lecturenumber - hepage : What's interesting about contract net?small lecturenumber - hepage : Distributed Planningsmall lecturenumber - hepage : Planning in Five Minutessmall lecturenumber - hepage : Planning in Five Minutessmall lecturenumber - hepage : Hierarchical Planningsmall lecturenumber - hepage : Distributed Planningsmall lecturenumber - hepage : Distributed Planningsmall lecturenumber - hepage : Distributed Planningsmall lecturenumber - hepage : Societies of agentssmall lecturenumber - hepage : Societies of agentssmall lecturenumber - hepage : Examplesmall lecturenumber - hepage : Rulessmall lecturenumber - hepage : Issuessmall lecturenumber - hepage : Summarysmall lecturenumber - hepage : Next timeDistributed Software DevelopmentCooperative MASChris BrooksDepartment of Computer ScienceUniversity of San FranciscoDepartment of Computer Science — University of San Francisco – p. 1/??22-2: Outline•What is an agent?•What are MAS/DAI problems?•Recommender Systems•Coordination◦Contract Net◦Distributed Planning•Societies of agentsDepartment of Computer Science — University of San Francisco – p. 2/??22-3: Distributed AI•Distributed AI (more often called Multiagent Systems thesedays) is the study of how multiple agents can and do interactwith each other.◦May include both human or computational agents.◦May be a closed or an open system◦Agents may be cooperative or self-interested.Department of Computer Science — University of San Francisco – p. 3/??22-4: Qualities of an agent•Autonomy•Adaptation•Goal-directed behavior•Has “beliefs” and “intentions”•Proactive•Situated within an environmentDepartment of Computer Science — University of San Francisco – p. 4/??22-5: Autonomy•Autonomy is a quality often attributed to agents.•An autonomous agent is able to rely on its percepts and pastexperience to make decisions, rather than asking a human forhelp.•This is a thorny area - most agents will not have completeautonomy.◦When might we not want an agent to have completeautonomy?•Challenge: Designing an agent that can reason about its ownautonomy and know when to ask for help.Department of Computer Science — University of San Francisco – p. 5/??22-6: Agent-oriented Programming•We can also think of agents as a programming paradigm.◦The next logical step after objects.◦“Objects do it for free, agents do it for money.”•Objects are receivers of actions, agents are actors.•It’s less useful to think of agent as an objective label than as asubjective description.•Agency is a useful abstraction for us as programmers.◦Allows us to think about a program at a higher level.•Treat something as an agent if that helps to understand,predict, or explain its behavior.◦Thermostats as agentsDepartment of Computer Science — University of San Francisco – p. 6/??22-7: Usefulness of the agent metaphor•Why bother with all this? We already know how to writeprograms.•Agents tend to be open-ended programs◦Difficult to specify in advance what they should do in allcases.•It’s helpful to talk about them as if they were intelligent.•“The robot wants to find the power supply.”•“The server believes that the client has reset.”•This assigning of mental states to programs is called theintentional stance.Department of Computer Science — University of San Francisco – p. 7/??22-8: Agents and Services•Service-oriented computing is a direct extension of the agentmetaphor•Idea: An agent is able to provide one or more services◦Specified at a high level, extracted from implementationaldetails.◦A service might actually involve gathering information fromother agents, delegating a task to another agent, orrecombining information.Department of Computer Science — University of San Francisco – p. 8/??22-9: Why solve DAI problems?•So a DAI or MAS problem involves a number of agentsinteracting with each other.•Why not construct a centralized system?◦Complexity◦Geographic or temporal separation of components◦Separate ownership◦Dynamic or changing systemDepartment of Computer Science — University of San Francisco – p. 9/??22-10: Some examples•NASA◦Coordinating Mars rovers◦multiple orbiting satellites•Scheduling◦Robots in factories◦Power generation and distribution◦Troop movement and deployment•Financial markets•P2P networks•etcDepartment of Computer Science — University of San Francisco – p. 10/??22-11: Recommender Systems•A problem that can be considered in a multiagent contextconcerns the gathering and combination of information frommultiple sources.•For example, Amazon, Netflix, MovieFinder, etc.•There are two different approaches to this problem:◦Content-based Filtering◦Collaborative FilteringDepartment of Computer Science — University of San Francisco – p. 11/??22-12: Content-based Filtering•In Content-based filtering, a model of each user is constructedbased on their expressed preferences.◦Things you bought, pages you looked at, things you saidyou liked.•A model of each product is constructed.•Those products that are “closest” to your preference model arerecommended.•Approach: Match users to
View Full Document
Unlocking...