lecture 04 intelligent agents

11

Click here to load reader

Upload: hema-kashyap

Post on 22-Jan-2018

56 views

Category:

Engineering


0 download

TRANSCRIPT

Page 1: Lecture 04 intelligent agents

Intelligent

AgentsLecture-4

Hema Kashyap

7 August 2015

Page 2: Lecture 04 intelligent agents

Agent

An Agent perceives itsenvironment through sensors

And acts upon the environmentthrough actuators or effectors

The complete set of input at agiven time is called perceptwhile can influence the actionsof an agent

An agent can be looked as asystem which has goals and inorder to implement them it mapspercept sequences into actions

Agents Performance measures:the success can be measured invarious ways i.e. in terms ofspeed, efficiency, accuracy,quality, power usage, money etc.

7 August 2015

Page 3: Lecture 04 intelligent agents

Ideal Rational Agent

For each possible percept sequence ,an ideal rational agentshould do whatever action is expected to maximize itsperformance measure, on the basis of the evidence providedby the percept sequence and whatever built in knowledge theagent has

An Agent behavior can be based on both its experience andbuilt in knowledge

A system is autonomous to the extent that its behavior isdetermined by its own experience.

Agent Program: A function that implements the agentsmapping from percept to actions. This will run on somecomputing device which we call the architecture. Eg.Software Agents (Softbots) to fly a flight simulator etc.

Agent =Architecture+ Program

7 August 2015

Page 4: Lecture 04 intelligent agents

Properties of Task Environments

1. Fully Observable /Partially Observable

2. Deterministic /Stochastic

3. Episodic/Sequential

4. Static/dynamic

5. Discrete/Continuous

6. Single Agent /Multi Agent

7. Cooperative

8. Strategic

7 August 2015

Page 5: Lecture 04 intelligent agents

Structure of Agent/Agent Architecture

1) Table Based Agents

2) Simple Reflex Agent/Percept Based Agents

3) Model Based Reflex Agents

4) Goal Based Agents

5) Utility Based Agents

6) Learning Agents

7 August 2015

Page 6: Lecture 04 intelligent agents

Table Based Agent

In Table based agent the action is looked up from a table

based on information about the agent’s percept

Its simply a mapping from percepts to action which is

defined by a program and implemented by a rule based

system , by neural network or by procedure

Drawbacks:

Table may be very large

learning table may take long time

Such system have little autonomy, as all actions are

predetermined

7 August 2015

Page 7: Lecture 04 intelligent agents

Simple Reflex Agent/Percept Based Agents

It acts according to a rule whosecondition matches the current state ,as defined by the percept

Such agents are designed based onthe condition action rule

First built a general purposeinterprets for condition action ruleand then to create rules set forspecific task environments

Eg. If (the car in front is braking)then (initiate braking)

They are also termed as Reactive orStimulus –Response agents

Features: Efficient

No internal representation forreasoning ,inference

No strategic planning

Percept based agents are not good formultiple opposing goals

7 August 2015

Page 8: Lecture 04 intelligent agents

Model Based Reflex Agents

It keeps track of thecurrent state of theworld using aninternal model. Itthen chooses anaction in the sameway as the reflexagent

It keeps track of thepart of the world itcan’t see now

7 August 2015

Page 9: Lecture 04 intelligent agents

Goal Based Agent Such agents keeps track

of the world state aswell as a set of goals itis trying to achieve andchooses an action thatwill eventually leads toachievement of its goals

Search and planning isneeded to achieve agentsgoal

Decision makinginvolves futureconsideration as goalbased

7 August 2015

Page 10: Lecture 04 intelligent agents

Utility Based Agents

If one world state is preferred t to another then it has higher utility for the agents

A utility function maps a state (a sequence of states ) onto a real number, whichdescribes associated degree of happiness.

A complete specification of the utility function allows rational decision in two kinds ofcases where goals are in adequate:

It uses a model of the world along with the utility function

That measures its preferences among state of the world. Conflicting Goals: i.e. only some of which can be achieved eg. Speed and safety

Several goals: none o of which can be achieve with certainty. In such cases likelihood ofsuccess depends upon the importance of goals

Then it chooses the actions that leads to the best expected utility where expected utilityis computed by averaging overall possible outcome states, weighted by the probabilityof the outcome.

7 August 2015

Page 11: Lecture 04 intelligent agents

Learning Agents

Learning allows the agent to operate in initially unknown environmentsand to become more competent than its initial knowledge alone mightallow

A learning agent can be divided into four conceptual components:

Learning element: responsible for making improvements

Performance Element: responsible for selecting external actions

Critic: the learning element uses feedback from the on how theperformance element should be modified, to do better in future

Problem Generator: responsible for suggesting actions that will lead tonew and informative experience

Eg. Taxi driving agents learns from feedback in form of reward andpenalty.

7 August 2015