New User Special Price Expires in

Let's log you in.

Sign in with Facebook


Don't have a StudySoup account? Create one here!


Create a StudySoup account

Be part of our community, it's free to join!

Sign up with Facebook


Create your account
By creating an account you agree to StudySoup's terms and conditions and privacy policy

Already have a StudySoup account? Login here

Artificial Intelligence

by: Earlene Cremin III

Artificial Intelligence CPSC 5185G

Earlene Cremin III

GPA 3.91


Almost Ready


These notes were just uploaded, and will be ready to view shortly.

Purchase these notes here, or revisit this page.

Either way, we'll remind you when they're ready :)

Preview These Notes for FREE

Get a free preview of these Notes, just enter your email below.

Unlock Preview
Unlock Preview

Preview these materials now for free

Why put in your email? Get access to more of this material and other relevant free materials for your school

View Preview

About this Document

Class Notes
25 ?




Popular in Course

Popular in ComputerScienence

This 4 page Class Notes was uploaded by Earlene Cremin III on Sunday October 11, 2015. The Class Notes belongs to CPSC 5185G at Columbus State University taught by Staff in Fall. Since its upload, it has received 52 views. For similar materials see /class/221209/cpsc-5185g-columbus-state-university in ComputerScienence at Columbus State University.


Reviews for Artificial Intelligence


Report this Material


What is Karma?


Karma is the currency of StudySoup.

You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!

Date Created: 10/11/15
Chapter 2 Intelligent Agents In discussing any topic it helps to have a central focus upon which to build our studies The authors of our textbook have elected to make the concept of rational intelligent agents their focal point There are some in the eld that claim that intelligent agents are quite passe belonging to another era in the manyfaceted study of this subject We however follow the lead of our authors and begin the discussion The de nition of an agent bears some resemblance to the de nition of an algorithm which we stated in an earlier lecture so we repeat that de nition here De nition An algorithm is a nite set of instructions which if followed will accomplish a particular task In addition every algorithm must satisfy the following criteria i input there are zero or more quantities which are externally supplied ii output at least one quantity is produced iii definiteness each instruction must be clear and unambiguous iv niteness if we trace out the instructions of the algorithm then for all valid cases the algorithm will terminate after a nite number of steps v e ctiveness every instruction must be suf ciently basic that it can in principle be carried out by a person using only a pencil and paper It is not enough that each operation be de nite as in iii but it also must be feasible or expressible as a nite sequence of instructions in the basic machine language of some computer Those who work in the area of engineering called Operations Research view algorithms within the context of optimization problems and add an additional requirement vi optimality the algorithm must always produce an optimal solution for problem instances with more than one optimal solution the algorithm may return any of these solutions We now turn to the de nition of an intelligent agent De nition An agent is anything that can be viewed as perceiving an environment through sensors and acting upon that environment through actuators Textbook Russell amp Norvig page 32 In this context a computer executing an algorithm can be viewed as a software agent perceiving its environment through its inputs keystrokes le contents and other sources and acting upon its environment through its output This approach is a bit too general for ready use in the study of algorithms where the more precise de nition of an algorithm is more useful nevertheless it does emphasize the valid point that the concept of agents has a strong basis in other topics of computer science In order to develop a language speci c to the study of intelligent agents we decide to use the term percept to refer to any of the agents inputs and the term percept sequence to denote the time sequence of percepts as input to the agent The agent may at any time act upon only the current percept or having memory may act upon the entire percept sequence beginning at a set time and continuing to the present Page 1 of5 CPSC 5185 Revised July 19 2010 Chapter 2 Intelligent Agents Strictly speaking we say that the agent s behavior is described by the agent function that maps a given percept sequence to an action or output This association may be via a table that lists the output associated with each percept sequence or a function that can be used to compute the output In principle the tabular approach is quite adequate In practice the tabular approach is use ll for only the smallest problems At one level we must keep these two approaches distinct in others we may ignore them assuming only that the output produced is appropriate to the percept sequence Our authors use a number of examples to illustrate the challenges found in developing an intelligent agent These are l the vacuumcleaner world and 2 the robotic taxi driver remember the robot taxi in the 1990 movie Total Recall We use the vacuumcleaner world to illustrate one important feature of intelligent agents and their specialization as search agents The world of this very simple problem has only two rooms arranged righttoleft When the robot is in one room its percepts are limited to what is in that room specifically whether it is clean or dirty If the robot moves from one room to another room it may store the condition of the previous room as a part of its percept sequence this example does not but it can no longer directly observe that room Assessing Agents the Rational Agent amp Its Performance Measure A rational agent is defined as one that does the right thing at all times that is it always acts in a way that will make it most successful according to a performance measure As the book notes it is often difficult to device a performance measure that states accurately the criteria for success of the agent too often these measures are only approximations of the true goals As the book notes rational behavior is not the same as behavior that always is successful in that the latter may require knowledge that is not available to the agent as any percept The book presents a definition of a rational agent For each possible percept sequence a rational agent should select an action that is expected to maximize its performance measure given the evidence provided by the percept sequence and whatever builtin knowledge the agent has Textbook Russell amp Norvig page 36 Note that a rational agent with a different performance measure might act differently Also note that a rational agent is not necessarily an omniscient agent Consider the problem of planning a taxi route This author has often taken a taxi in a crowded city and been asked whether he wanted the cheapest route or the fastest route possibly including a toll road Note that we here have two different performance measures that are likely to lead to different route plans Note also that an omniscient agent knowing that a big truck is about to jackknife and block the faster route for about two hours would apply different criteria Backing off the requirement of omniscience for an agent yields the benefit that we begin to speak of agents that we actually can design and possibly improve later For example a rational agent that does route planning might be improved by providing a facility to monitor current traffic conditions and alter routes in response to accidents and the like At present there are a number of services such as XM Satellite Radio that provide traffic information but these are limited to larger cities and not very detailed Page 2 of5 CPSC 5185 Revised July 19 2010 Chapter 2 Intelligent Agents Two additional features desired of rational agents are learning and autonomy Each desired feature has a simple de nition that will suffice for our discussions We would like a rational intelligent agent to be able to learn in that it devises new sequences of actions for a percept sequence for which its previous sequence of actions have lead to disappointing results This involves both reinforcing of action sequences that lead to positive results and deprecating those sequences that do not We would also like the agent to be able to modify its actions in response to observations of the environment thus the robot taxi driver would note the presence of snow and ice on a road and devise a different plan for braking the taxi Complete autonomy is rarely sought in an agent even those for which detailed control is almost an impossibility Consider a space probe in orbit around Saturn Under the most ideal circumstances the control loop from the Earth for this probe is over two hours that is if the probe measures something and radios home it takes more than an hour for the signal to reach the earth and another hour for the return signal with instructions to reach the probe We must have some degree of autonomy for the probe expecting it to function according to specific highlevel strategies with the possibility of a safe mode in which the probe sends data and then executes a completely specified hibernation program awaiting instructions The Agent s Environment An intelligent agent is expected to interact with an external environment The environment can vary from something as simple as the external data set processed by an implementation of a simple algorithm to a dynamic unpredictable and constantly changing scene through which the robot must navigate For the latter think of a space probe attempting to traverse the rings of Saturn where none of the boulders that comprise the ring have been mapped with sufficient accuracy to allow navigation by maps It is within this context that the textbook presents the automatic taxi driver as a sample of a proposed rational intelligent agent We should take this example as a topic for discussion in which we can enumerate and debate the issues that must be solved before attempting such a project However we should avoid the temptation to make jokes about human taxi drivers There are a number of properties of the task environment that we should consider when attempting the design of an agent to operate in such an environment We list them here and comment on the textbook s discussion of these properties Fullyiobservable vs Partiallyiobservable Simply put can the agent see everything that needs to be seen We do not require that the environment be totally observed at once consider sitting in a room in which you are facing one way and not observing what is behind your head You can turn your head should the need arise and quickly observe what is not currently observed so we call this observable We may have noisy data incomplete data or regions of the environment that are not yet directly observable but which can be explored and observed later Admittedly the last case is a special case of that with incomplete data but we list it separately Page 3 of5 CPSC 5185 Revised July 19 2010 Chapter 2 Intelligent Agents The author of these notes will add an example from his experience working in the Star Wars program 7 defense against ballistic missiles The warhead hydrogen bomb of the ballistic missile is contained in a reientry vehicle called an RV 7 not to be confused with a recreational vehicle such as a Winnebago and the goal of the defense was to track and intercept the RV thus destroying it and its bomb An early experiment in this technology was called HOE Homing Overlay Experiment The HOE vehicle used an infrared telescope to track its target and guide the interceptor vehicle to a point in space at which the interception would be attempted The task of devising the guidance algorithm was given to an engineer with many years of experience in designing such algorithms for radariguided airitoiair missiles The guidance algorithm he devised called for a number of inputs describing the position of the target as observed by the interceptor two angles called azimuth and elevation the range to the target and the range rate how fast the target and the interceptor were approaching each other Radars can measure both the range to a target via measuring the time delay of the return pulse and the range rate by observing the Doppler shift of the return signal Telescopes either optical or infrared can measure neither so the wonderful algorithm was useless Deterministic Stochastic and Apparently Stochastic Some simple environments mostly those arti cial environments for research or teaching can be considered to be deterministic in that the future state of the environment can be determined completely by application of known laws and formulae to the percept sequence representing the measurements of the current state of the environment Realistic examples of such environments include the orbits of arti cial satellites around the Earth in which case the overwhelming in uence is the well understood gravity of the planet with secondary in uences from the equally well understood drag of the air remaining at that altitude A more interesting example would be the movement of water over Niagara Falls In principle this is completely deterministic in that every water molecule follows motions specified by Newton s laws As a matter of fact computation of the water motion using these basic laws would require considerably more computational power than is currently available on the planet and would require considerably longer than a human life to complete Other approaches based on statistical dynamics and the associated laws of uid ow are inherently stochastic in nature and much more applicable We then define a deterministic environment not as one following well known physical laws but one that can be predicted using these laws in a timely fashion using only the computational power available to the agent We also exclude from this classification all those probably deterministic environments the behavior of which cannot be predicted from the available percept sequence In other words many environments are best modeled as stochastic even though they are not actually such Three categories of environments as defined by this textbook are l Deterministic in which the future state of the environment may reasonably be predicted by the agent using the computational power available 2 Strategic in which the future state of the environment is determined by other agents presumed to be deterministic and not random 3 Stochastic in which the future state of the environment cannot be predicted Page 4 of5 CPSC 5185 Revised July 19 2010


Buy Material

Are you sure you want to buy this material for

25 Karma

Buy Material

BOOM! Enjoy Your Free Notes!

We've added these Notes to your profile, click here to view them now.


You're already Subscribed!

Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'

Why people love StudySoup

Jim McGreen Ohio University

"Knowing I can count on the Elite Notetaker in my class allows me to focus on what the professor is saying instead of just scribbling notes the whole time and falling behind."

Janice Dongeun University of Washington

"I used the money I made selling my notes & study guides to pay for spring break in Olympia, Washington...which was Sweet!"

Bentley McCaw University of Florida

"I was shooting for a perfect 4.0 GPA this semester. Having StudySoup as a study aid was critical to helping me achieve my goal...and I nailed it!"

Parker Thompson 500 Startups

"It's a great way for students to improve their educational experience and it seemed like a product that everybody wants, so all the people participating are winning."

Become an Elite Notetaker and start selling your notes online!

Refund Policy


All subscriptions to StudySoup are paid in full at the time of subscribing. To change your credit card information or to cancel your subscription, go to "Edit Settings". All credit card information will be available there. If you should decide to cancel your subscription, it will continue to be valid until the next payment period, as all payments for the current period were made in advance. For special circumstances, please email


StudySoup has more than 1 million course-specific study resources to help students study smarter. If you’re having trouble finding what you’re looking for, our customer support team can help you find what you need! Feel free to contact them here:

Recurring Subscriptions: If you have canceled your recurring subscription on the day of renewal and have not downloaded any documents, you may request a refund by submitting an email to

Satisfaction Guarantee: If you’re not satisfied with your subscription, you can contact us for further help. Contact must be made within 3 business days of your subscription purchase and your refund request will be subject for review.

Please Note: Refunds can never be provided more than 30 days after the initial purchase date regardless of your activity on the site.