Agent System

8/14/2019 Agent System

1/65

Artificial

IntelligenceByNoman Hasany

Assistant Professor

Computer Engineering Department

SIR SYED UNIVERSITY OF ENGINEERING & TECHNOLOGY, KARACHI


2/65

Course Outline

Theory

AI (Intro)

Searching

Intelligent Agents Logic and Inference

Knowledge-based

Systems

Natural LanguageProcessing

Neural Nets

Application

Prolog

Natural Language Tool

kit (NLTK) + Python


3/65

AI


4/65

Heuristic

Heuristic is a rule or piece of information thatis used to make a problem-solving method moreeffective (to achieve the goal) ormore efficient(fast and/or have less space) .

Example: a person who is looking for a suitablegift would go to straight to the shop that heconsidered to be most likely to have a suitablegift, then to the next most likely and so on.

Humans use such heuristics all the time tosolve all kinds of problemsComputers canalso use.


5/65

Why study AI?

Search engines

Labor

Science

Medicine/Diagnosis

Appliances What else?


6/65

AI State of the art

Deep Blue defeated the reigning world chess championGarryKasparov in 1997

Automated Reasoning methods proved a mathematicalconjecture (Robbins conjecture) unsolved for decades

No hands across America (driving autonomously 98% ofthe time from Pittsburgh to San Diego) Military still the strongest factor for AI research: During

the 1991 Gulf War, US forces deployed an AI logisticsplanning and scheduling program that involved up to50,000 vehicles, cargo, and people

NASA's on-board autonomous planning programcontrolled the scheduling of operations for a spacecraft

Proverb solves crossword puzzles better than mosthumans


7/65

AI State of the art

Have the following been achieved by AI? World-class chess playing

Playing table tennis

Cross-country driving Solving mathematical problems

Engage in a meaningful conversation

Handwriting recognition Observe and understand human emotions

Express emotions


8/65

AI (John McCarthy)

It is the science and engineering of

making intelligent machines, especially

intelligent computer programs. It is related

to the similar task of using computers to

understand human intelligence, but AI

does not have to confine itself to methods

that are biologically observable.


9/65

AI (Marvin Minsky)

Artificial Intelligence is the science of

making machines do things that require

intelligence if done by men."


10/65

AI

Views of AI fall into four categories: Rationally means logically


11/65

What is AI?

The exciting new effort tomake computers thinks machine with minds, in thefull and literal sense(Haugeland 1985)

The art of creatingmachines that performfunctions that requireintelligence when

performed by people(Kurzweil, 1990)

The study of mental facultiesthrough the use ofcomputational models(Charniak et al. 1985)

A field of study that seeks toexplain and emulateintelligent behavior in termsof computational processes

(Schalkol, 1990)Systems that think likehumans

Systems that thinkrationally

Systems that act likehumans

Systems that act rationally

[Above are Emulation]

(Imitate)

[Above are Simulation]

(Replicate)


12/65

Acting humanly: Turing Test

Turing (1950) "Computing machinery and

intelligence":

operational test for intelligent behavior: the

Imitation Game

components needed for passing the test:

natural language processing, knowledge

representation, automated reasoning,machine learning


13/65

Acting humanly: Turing Test


14/65

Thinking humanly: cognitive

modeling

not only the results, but also the reasoning steps

made for solving a problem must be similar

requires scientific theories of internal activities of

the brain: introspection (the observation of thingsinternal to one's self)and psychologicalexperiments

cognitive science: computer models from AI andexperimental techniques from psychology

-theories of the workings of the human mind


15/65

Thinking rationally:

"laws of thought"

Aristotle: what are correct arguments/thoughtprocesses? syllogisms: patterns for argument structures that always

yielded correct conclusions when given correct premises

several Greek schools developed various forms oflogic: notation and rules of derivation for thoughts;

direct line through mathematics and philosophy tomodern AI

problems: 1.difficulties in translating everyday knowledge into a

logical formalism (the presence of uncertainty)

2.scalability (degree/range) of logic-based reasoning

systems


16/65

Acting rationally: rational agent

Rational behavior: doing the right thing

The right thing: that which is expected to maximize goalachievement, given the available information

doesn't necessarily involve thinking

but thinking should be in the service of rationalaction (thought is involve in design phase)

more amenable to scientific development than human-like behavior or thought approaches


17/65

Rational agents

An agent is an entity that perceives and acts Abstractly, an agent is a function from percept

histories to actions:

For any given class of environments and tasks,we seek the agent (or class of agents) with thebest performance

Caveat (limitation): computational limitationsmake perfect rationality unachievable design best program for given machine resources


18/65

Intelligence

Intelligence is the computational part ofthe ability to achieve goals in the world.

Varying kinds and degrees of intelligence

occur in people, many animals and somemachines.

Ultimate intelligence is the spiritual part of

the ability to achieve goals in this worldand for the world hereafter. (rabbana atinafidduniya hasana wafil akhirate hasana)


19/65

Types of AI (MSN Encarta) Symbolic AI: Symbolic AI is based in logic. It uses sequences of

rules to tell the computer what to do next. Expert systems consistof many so-called IF-THEN rules: IF this is the case, THEN do that.Since both sides of the rule can be defined in complex ways, rule-based programs can be very powerful.

Connectionist AI: Connectionism is inspired by the brain. It is

closely related to computational neuroscience, which modelsactual brain cells and neural circuits. Connectionist AI usesartificial neural networks made of many units working in parallel.Each unit is connected to its neighbors by links that can raise orlower the likelihood that the neighbor unit will fire (excitatory andinhibitory connections respectively). Neural networks that are able

to learn do so by changing the strengths of these links, dependingon past experience. These simple units are much less complexthan real neurons. Each can do only one thing: for instance, reporta tiny vertical line at a particular place in an image. What matters isnot what any individual unit is doing, but the overall activity-patternof the whole network.


20/65

Types of AI (MSN Encarta)

Evolutionary AI: Evolutionary AI draws onbiology. Its programs make random changes intheir own rules, and select the best daughterprograms to breed the next generation. This

method develops problem-solving programs,and can evolve the brains and eyes of robots.It is often used in modeling artificial life (A-Life).A-Life studies self-organization: how order arisesfrom something that is ordered to a lesser

degree. Biological examples include theflocking patterns of birds and the developmentof embryos.


21/65

Branches of AI (John McCarthy)

Logical AI: What a program knows about the world ingeneral the facts of the specific situation in which it mustact, and its goals are all represented by sentences ofsome mathematical logical language. The program

decides what to do by inferring that certain actions areappropriate for achieving its goals.

Search: AI programs often examine large numbers ofpossibilities, e.g. moves in a chess game or inferencesby a theorem proving program. Discoveries are

continually made about how to do this more efficiently invarious domains.


22/65


Pattern recognition: When a program

makes observations of some kind, it is

often programmed to compare what it

sees with a pattern. For example, a visionprogram may try to match a pattern of

eyes and a nose in a scene in order to find

a face, for finding word sense, tag sensein a natural language text, for playing next

move in a chess game


23/65

Representation: Facts about the world have to berepresented in some way. Usually languages ofmathematical logic are used.

Inference: From some facts, others can be inferred.Mathematical logical deduction is adequate for some

purposes Common sense knowledge and reasoning: This is thearea in which AI is farthest from human-level, in spite of thefact that it has been an active research area since the1950s. While there has been considerable progress, e.g. in

developing systems of non-monotonic reasoning andtheories of action, yet more new ideas are needed. The Cycsystem contains a large but spotty collection of commonsense facts.



24/65

Learning from experience: The approaches to AIbased on connectionism and neural nets specialize inthat. There is also learning of laws expressed in logic.Programs can only learn what facts or behaviors theirformalisms can represent, and unfortunately learning

systems are almost all based on very limited abilities torepresent information.

Planning: Planning programs start with general factsabout the world, facts about the particular situation and astatement of a goal. From these, they generate a

strategy for achieving the goal. In the most commoncases, the strategy is just a sequence of actions. Epistemology: This is a study of the kinds of knowledge

that are required for solving problems in the world.



25/65

Ontology: Ontology is the study of the kinds of things thatexist. In AI, the programs and sentences deal with variouskinds of objects, and we study what these kinds are andwhat their basic properties are. Emphasis on ontologybegins in the 1990s.

Heuristics: A heuristic is a way of trying to discoversomething or an idea embedded in a program. Heuristicfunctions are used in some approaches to search tomeasure how far a node in a search tree seems to be froma goal. Heuristic predicates that compare two nodes in asearch tree to see if one is better than the other, i.e.constitutes an advance toward the goal,

Genetic programming: Genetic programming is atechnique for getting programs to solve a task by mating(friendly) random Lisp programs and selecting fittest inmillions of generations.



26/65

Foundations of AI

Philosophy

Mathematics

Economics

Neuroscience

Psychology

Computer engineering

Control theory

Linguistics


27/65

Engineering Intelligent Entities

The science of understanding intelligent

entities by developing theories that attempt to

explain and predict the nature of such entities.

The engineering of intelligent entities.


28/65

Engineering Intelligent Entities

The engineering of such intelligent entities means thatwe want to build machines to perform tasks for us.

However, these machines must act in an intelligent way.

Intelligence can be considered to be manifest in anumber of different but interconnected forms ofbehaviour, such as: reasoning about beliefs,

reasoning about actions, the ability to communicate, the ability to learn, both from mistakes and also through gaining

new knowledge, social awareness.


29/65

Agents

The key notion here is that we want to equip such intelligentsystems, known as agents, with the ability to think and actautonomously.

There is actually no universal agreed upon definition of an agent buta popular definition has been given by Wooldridge and Jennings(1995)

An agentis a computer system that is situatedin someenvironment, and that is capable ofautonomous actionin this environment in order to meet its designobjectives.

In addition to this, a collection of such agents situated together in anenvironment and capable of interacting in with one another is knownas a multi-agent system.


30/65

What is an (Intelligent) Agent?

An over-used, over-loaded, and

misused term.

Anything that can be viewedas

perceiving its environment through

sensors and acting upon thatenvironment through its effectors to

maximize progress towards its goals.


31/65

What is an (Intelligent) Agent?

PAGE (Percepts, Actions, Goals,Environment)

Task-specific & specialized: well-defined

goals and environment The notion of an agent is meant to be a

tool for analyzing systems,

It is not a different hardware or newprogramming languages


32/65

Example: Human mind as network of thousands or millionsof agents working in parallel. To produce real artificial

intelligence, this school holds, we should build computer

systems that also contain many agents and systems for

arbitrating among the agents' competing results.

Distributed decision-making

and control

Challenges:

Action selection: What next action

to choose

Conflict resolution

Intelligent Agents and Artificial

Intelligence

sen

sors

effectors

Agency


33/65

AgentsHuman agent: eyes, ears, and other organs for

sensors;

hands, legs, mouth, and other body parts foractuators

Robotic agent: cameras and infrared range

finders for sensors;

various motors for actuators


34/65

Intelligent Agents

In order for these entities to be deemed as intelligent, there are anumber of capabilities that we would expect such agents topossess, again taken from a list defined by Wooldridge andJennings (1995)

Reactivity: Intelligent agents are able to perceive their environment,and respond in a timely fashion to changes that occur in it in orderto satisfy their design objectives.

Proactiveness: Intelligent agents are able to exhibit goal-directedbehaviour by taking the initiative in order to satisfy their designobjectives.

Social ability: Intelligent agents are capable of interacting with otheragents (and possibly humans) in order to satisfy their designobjectives.


35/65

Intelligent Agents

Environment

Sensor inputAction output


36/65

A Windshield Wiper Agent

How do we design a agent that can wipe the windshields

when needed?

Goals? Percepts?

Sensors?

Effectors?

Actions?

Environment?


37/65

A Windshield Wiper Agent

Goals: Keep windshields clean &maintain visibility

Percepts: Raining, Dirty

Sensors: Camera (moist sensor)

Effectors: Wipers (left, right, back)

Actions: Off, Slow, Medium, Fast

Environment: Inner city, freeways,

highways, weather


38/65

Interacting Agents

Collision Avoidance Agent (CAA)

Goals: Avoid running into obstacles

Percepts ? Sensors?

Effectors ?

Actions ? Environment: Freeway


39/65

Interacting Agents

Lane Keeping Agent (LKA)

Goals: Stay in current lane

Percepts ? Sensors?

Effectors ?

Actions ? Environment: Freeway


40/65

Interacting Agents

Collision Avoidance Agent (CAA)

Goals: Avoid running into obstacles

Percepts: Obstacle distance, velocity, trajectory

(route) Sensors: Vision, proximity (closeness)sensing

Effectors: Steering Wheel, Accelerator,

Brakes, Horn, Headlights Actions: Steer, speed up, brake, blow horn,

signal (headlights)

Environment: Freeway


41/65

Interacting Agents

Lane Keeping Agent (LKA)

Goals: Stay in current lane

Percepts: Lane center, laneboundaries

Sensors: Vision

Effectors: Steering Wheel,

Accelerator, Brakes Actions: Steer, speed up, brake

Environment: Freeway


42/65

Structure of Intelligent Agents

Agent = architecture + program

Agent program: the implementation off: P* A, theagents perception-action mapping

function Skeleton-Agent(Percept) returnsAction

memory UpdateMemory(memory, Percept)

Action ChooseBestAction(memory)

memory UpdateMemory(memory, Action)

returnAction

Architecture: a device that can execute the agent

program (e.g., general-purpose computer, specialized

device, beobot, etc.)


43/65

Beobot

What is a Beobot?

It is an autonomous robot that uses off-the-

shelf PC hardware and software to perform its

computation

It is designed to run in actual outdoor

unconstrainedenvironments, which is a

far cry from the insulated test labs thatrobots are mainly tested in.

U i l k t bl t d


44/65

Using a look-up-table to encode

f: P* A Example: Collision Avoidance

Sensors: 3 proximity sensors

Effectors:Steering Wheel, Brakes

How to generate?

How large?

How to select action?

agent

obstacle

sensors

U i l k t bl t d


45/65

Using a look-up-table to encode

f: P* A

Example: Collision Avoidance

Sensors: 3 proximity sensors

Effectors:Steering Wheel, Brakes

How to generate: for each p Pl Pm Prgenerate an appropriate action, a S B

How large: size of table = #possible percepts times # possibleactions = |Pl | |Pm| |Pr| |S| |B|

E.g., P = {close, medium, far}3

A = {left, straight, right} {on, off}then size of table = 27*3*2 = 162

How to select action? Search.

agent

obstaclesensors


46/65

Information agents Manage the explosive growth of information.

Manipulate or collate information from many distributed sources. Information agents can be mobile or static.

Examples:

BargainFindercomparison shops among Internet stores for CDs

FIDO the Shopping Doggie (out of service)

Internet Softbot infers which internet facilities (finger, ftp, gopher)

to use and when from high-level search requests.

Challenge: ontologies for annotating Web pages (eg, SHOE).
http://www.bargainvillage.com/bargainFinder/software.htmhttp://www.shopfido.com/http://www.cs.washington.edu/research/projects/softbots/www/internet-softbot.htmlhttp://www.cs.washington.edu/research/projects/softbots/www/internet-softbot.htmlhttp://www.shopfido.com/http://www.bargainvillage.com/bargainFinder/software.htm


47/65

Intelligent Agents

Agents observe the environment in which they aresituated and perform actions based on their knowledgeof their environment.

However, observing an environment is no easy task for anumber of reasons: the agent needs be equipped with some suitable representation

of the environment which it inhabits the agent needs to be able to reason about the state of the

environment

the agent needs to be able to update its beliefs when theenvironment changes the agent needs to know how its action will affect the

environment it inhabits.


48/65

Environments Russell and Norvig (1995) have given a

classification of the different types of environmentthat an agent can inhabit:

Accessible vs inaccessible

Deterministic vs non-deterministic Episodic vs non-episodic Static vs dynamic Discrete vs continuous

We will look at a brief description of each


49/65

Accessible vs. Inaccessible

An accessible environment is one in which theagent can obtain complete, accurate, up-to-dateinformation about the environments state.

Most moderately complex environments(including, for example, the everyday physicalworld and the Internet) are inaccessible.

The more accessible an environment is, thesimpler it is to build agents to operate in it.

D t i i ti N


50/65

Deterministic vs. Non-

Deterministic

Output problem A deterministic environment is one in which any

action has a single guaranteed effect there isno uncertainty about the state that will result

from performing an action.

The physical world can to all intents andpurposes be regarded as non-deterministic.

Non-deterministic environments present greaterproblems for the agent designer.


51/65

Episodic vs. Non-Episodic

Input Problem In an episodic environment, the performance of an agent is

dependent on a number of discrete episodes, with no link betweenthe performance of an agent in different scenarios.

Episodic environments are simpler from the agent developersperspective because the agent can decide what action to performbased only on the current episode it need not reason about theinteractions between this and future episodes.

An example of an episodic environment is a mail sorting system.


52/65

Static vs. Dynamic

A static environment is one that can be assumedto remain unchanged except by the performanceof actions by the agent.

A dynamic environment is one that has otherprocesses operating on it, and which hencechanges in ways beyond the agents control.

The physical world is a highly dynamicenvironment.


53/65

Discrete vs. Continuous

An environment is discrete if there are a fixed,finite number of actions and percepts in it.

A chess game is an example of a discreteenvironment, and taxi driving is an example of acontinuous one.

This is not really an intrinsic property of theenvironment, but more a property of how wechoose to model the environment. So, you canthink of your perceptions of the world or youractions as being discrete or continuous.


54/65

Environments

So, the type of environment that an agentinhabits effects its design and behaviour.

In order for an agent to think rationally itneeds to be aware of the environment itinhabits, which requires knowledge

representation, and it needs to be able toreason about this knowledge in order toact rationally and fulfil its tasks.


55/65

Multi-Agent Systems

However, the situation becomes even more complicatedwhen there are other such agents operating within thesame environment.

Now, the agent must be able to not only reason about itsown beliefs but also about the beliefs of other agents

and

it also needs to reason about the way the environmentcan be changed by these other agents.

All this requires rational reasoning.


56/65

Reasoning about Beliefs

In order to represent agents environments in a mannerthat they can understand we use logic.

We encode information about the environment in the

form of logic and this information forms the agentsknowledge base and thus its beliefs about the world.

In order to enable an agent to reason with thisknowledge, the agent needs to have a set of rules thattell it how this environment can change.


57/65

Reasoning about Action

The agents knowledge about the environment shouldalso tell it how the environment will change when itperforms actions.

So the ability to act rationally is intrinsically (built-in)linked with the ability to think rationally.

The agent will have specific goals that it is trying toachieve and these goals are represented as particularstates of the environment.

In order to act rationally the agent must choose an actionto execute that will achieve this particular goal.


58/65

DomainApplications

Agents spheres of knowledge will differ

depending upon the particular domain

they are constructed for use in.

For example


59/65

Example Domains

The following domains are examples of

where agent technology is being

deployed:

The internet: e.g. softbots (software robots)

which visit websites and gather information

from them. e.g. for comparison shopping, or for harvesting

information.


60/65

Example Domains

Medicine: multi-agent systems can be used

for simulation, decision making,

communication and risk assessment

purposes one such group working on the application of multi-agent systems to medicine is the Advanced

Computation Lab at Cancer Research UK


61/65

Example Domains

Air traffic control: agents are deployed insystems to assist in the control of air trafficand the provision of information. example: the OASIS system tested at Sydney

airport.

Spacecraft: agents are currently being testedfor use in spacecraft computers e.g. for autonomous navigation and detection of

errors/malfunctions.

e.g. NASAs Deep Space 1 mission.


62/65

Example Domains

Telecommunications systems: e.g. for managing

dynamic information concerning the routing of

network traffic.

Law: to support decision making and information

management involved in legal reasoning.

These are just a few domains that are currently

making use of agent technology and KR&R methods.

There are many more application domains.


63/65

Agents and KR&R

All the previous application domains for agent systemseach have their own aims with respect to the problemsthat they address.

So, the knowledge and reasoning embodied will differdepending upon the domain.

But, what is common to all the application areas thatmake use of AI and agent technology is that they needsystems that possess the relevant knowledge of thedomain and environment and they also need to be ableto reason effectively with this information in order to fulfiltheir specific tasks.


64/65

Agents and KR&R

Expressive logics, such as modal and descriptionlogics, are being used as an efficient and effectiveway of dealing with KR&R (Knowledgerepresentation and reasoning) in intelligentsystems, as we will come to see during the course.

Further detailed information about agent technologyand applications can be found in:

M. J. WooldridgeAn Introduction to Multi-Agent Systems

John Wiley and Sons Publishers


65/65

Summary

Overview of:

AI

Intelligent agents

Environments

Reasoning about beliefs and actions

Domain applications

Documents

Agent System