Pengantar Kecerdasan Buatan Intelligent Agent 2 Why we

Pengantar Kecerdasan Buatan Intelligent Agent

2 Why we need agent? • AI can not be functioned or useless without an entity or an object that can act intelligently • An entity that is designed to act using AI is called as an agent

3 Agent Definition • An Agent is anything that can be viewed as perceiving its environments through sensors and acting upon that environments through actuators Agent = Architecture + Program

4 Agent Definition (Cont’d) Human Agent • Sensors: • • • Eyes Ears Tongue Actuators: • • Robotic Agent Hands Legs Sensors: • • • Cameras Infrared GPS Actuators: • • Software Agent Motors Wheels Sensors: • • Keystrokes Files Networks Actuators: • • Screen Display Write Files

5 Agent Anatomy An agent should consist of : Sensors Actions The ENVIRONMENT Actuators Percepts

6 Illustration of an Agent Sensors ? Actuators

7 Example : A Vacuum Cleaner Agent

8 Good Behavior: The Concept of Rationality • A rational agent is one that does the right thing. • What does it mean to do the right thing? • • The right action is the one that will cause the agent to be most successful We will need some way to measure success

9 Performance Measurements • There is no single standard which can be applied to all types of agent • Therefore, we will need to create an objective performance measure • As a general rule, it is better to design performance measures according to what one actually wants in the environment, rather that according to how one thinks the agent should behave

10 Rationality • What is rational at any given time depends on four things The performance measure that defines the criterion on success • The agent’s prior knowledge of the environment • The agent’s sense sequences from the environment • The actions that the agent can perform •

11 A Rational Agent For each possible percept (sense) 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 in the environment

12 Omniscience • Omniscience: Knowledge of all things • An omniscient agent knows the actual outcome of its actions and can act accordingly. • In reality omniscience is impossible • Rationality is not the same as perfection • Rationality maximizes expected performance, while perfection maximizes actual performance

13 Learning • A rational agent not only to gather information but also to learn as much as possible from what it perceives • The agent’s initial configuration should reflect some prior knowledge of the environment, but as the agent gains experience this may be modified and augmented

14 Autonomous • A rational agent should be autonomous • It should learn what it can to compensate for partial or incorrect prior knowledge

15 Mapping • We can describe any particular agent by making a table of action it takes in response to each possible percept sequence. Such list is called a mapping from percept sequences to action. • Ideal mapping : specifying which action an agent ought to take in response to any given percept sequence provides a design for an ideal agent.

16 Why not just look up the answers ? • The table needed for something as simple as an agent that can only play chess would be about 35100. . • It would take quite a long time for the designer to build the tables. • The agent has no autonomy at all. Because the calculation of best action is entirely built-in. So if the environment changed in some unexpected way, the agent would be lost!

17 The Nature of Environments • We must think about task environments, which are essentially the “problems” to which rational agents are the “solutions” • PEAS: Performance measure, Environment, Actuators, Sensors • In designing an agent, the first step must always be to specify the task environment as fully as possible

Example of Agents and PEAS Agent Type Performance Measure(Goals) Environment 18 Actuators (Actions) Sensors (Percepts) Roads, traffic, pedestrians, customers Steering, accelerators, brake, signal, horn, display Cameras, sonar, speedometer, GPS, engine sensors, odometer, fee meter Keyboard entry symptoms, patient answers, history Taxi driver Safe, fast, legal, comfort, max. profits Medical diagnosis system Healthy patient, min. cost, lawsuits Patient, hospital, staff Display, questions, tests, diagnoses, treatments Satellite image analysis system Correct image categorization Downlink from orbiting satellite Display categorization of scene Color pixel arrays Part-picking robot Percentage parts in correct bins Conveyor belt with parts, bins Jointed arm and hand Camera, joint angle sensors Refinery control Max. purity, yield, safety Refinery operators Valves, pumps, heaters, displays Temperature, pressure, chemical sensors Interactive English tutor Max. student’s score on test Set of students, testing agency Display exercises, suggestions, corrections Keyboard entry

19 Properties of Task Environment • Fully observable VS partially observable • • An environment is fully observable if the sensors detect all aspects that are relevant to the choice of action. Deterministic VS stochastic • If the next state of the environment is completely determined by the current state and the action executed by the agent, then the environment is deterministic.

20 Properties of Task Environment (Cont) • Discrete VS continuous • • The discrete/continuous distinction can be applied to the state of the environment, to the way time is handled, and to the percepts and actions of the agent. Single agent VS multi-agent • If the environment can change because of other entities then it is a multi-agent environment.

21 Properties of Task Environment (Cont) • Episodic VS sequential • • An Episode consists of the agent perceiving and then performing a single action, the next episode doesn’t depend on the actions taken in previous episode. Static vs. dynamic • If the environment can change while an agent is deliberating, then the environment is dynamic.

22 Task environment and Their Characteristics

23 Four Basic Kinds of Agent Program • Simple reflex agents • Model-based reflex agents • Goal-based agents • Utility-based agents

24 Simple Reflex Agents • Use a “condition-action rule” to complete a task • • E. g. : If car-in-front-is-braking then initiate-braking Very similar to the human reflex

25 Simple Reflex Agents (Cont) Sensors What the world is like now Condition-action rule What action I should do now Actuators

26 Model-based Reflex Agents • Relying on reflex is not enough ? ? • Agents need to have Information on how the world evolves independently of the agent • Information about how the agent’s own action affect the world. •

27 Model-based Reflex Agents (Cont) States How the world evolves Sensors What the world is like now What my action do Condition-action rule What action I should do now Actuators

28 Goal-based Agents • Having an agent that only respond to impulse is not something that we wanted to do. • Agent needs some sort of goal information, which describe what condition is desirable.

29 Goal-based Agents (Cont) States How the world evolves What my action do Goals Sensors What the world is like now What it will be if I do… What action I should do now Actuators

30 Utility-based Agents • Achieving a goal alone at all cost is not the purpose of intelligent agent. • You can achieve goal in different ways, varying in efficiency and affectivity • Agents need more than “Happy” or “Unhappy” states. • Utility is a function that maps a state which describes the associated degree of happiness.

31 Utility-based Agents (Cont) States How the world evolves What my action do Utility Sensors What the world is like now What it will be if I do… How happy I will be in Such state What action I should do now Actuators

32 Learning Agents • Agents needs to come into being • Learning allows agents to operate in initially unknown environments and to become more competent than its initial knowledge alone might allow

33 Learning Agents (Cont) Performance standard Sensors Critic Feedback Learning Element Learning goals Problem Generator Changes Performance element Knowledge Actuators

34 How confidence are you about AI? • Turing Test http: //cogsci. ucsd. edu/~asaygin/tt/ttest. html • Berikan pandangan Anda mengenai Turing test, apakah sudah ada alat /program komputer, yang benar-benar sudah bisa menipu manusia? Jelaskan alasan Anda • Coba lakukan percakapan dengan chat bot, misal: Alice. Rekam percakapan Anda dalam bentuk tabel sbb: No. Anda Chat Bot Komentar (tujuan percakapan) Adakah bagian yang menurut Anda menjelaskan mengenai kecerdasan? • Alice termasuk tipe agent apa (reflex, model, goal, utility)? Mengapa? • Bagaimana pembentukan lingkungannya (slide #22) •

35 Kuis 1 1. Definisikan kecerdasan buatan menurut Anda. 2. Gambarkan 4 jendela pendekatan dalam AI dan jelaskan artinya. 3. Berikan contoh beberapa aplikasi yang menurut Anda menerpkan konsep AI, dan berikan bagaimana AI berperan dalam aplikasi tersebut?