AI Course with Sebastian Thrun and Peter Norvig: Udacity Course: 01

“Artificial Intelligence Course was made in 2011 by Peter Norvig and myself at Stanford University and became basically the first very large massive open online course 160,000 students at the time then sparked the development of many many other such courses. The course is comprised of many units covering all the bases in artificial intelligence. It’s modelled after the very successful book by Stuart Russel and Peter Norvig which is the number one book on artificial intelligence and I really hope you get to enjoy this class it’s the original class diving.”  Sebastian THRUN, Co-FOUNDER&CEO, UDACITY

Artificial Intelligence: A Modern Approach (AIMA) is a university textbook on artificial intelligence, written by Stuart J. Russell and Peter Norvig.

  
Welcome to the Artificial Intelligence class on Udacity. Sebastian Thrun and Peter Norvig are your teachers. You can take this class at your own pace. You need to match up the quizzes, the homework assignments, the midterm and the final exam to succeed in this course.

THE PURPOSE OF THIS CLASS :
1. TO TEACH YOU THE BASICS OF ARTIFICIAL INTELLIGENCE (so you’ll be able to talk to people in the field and understand the basic tools of the trade)
2. TO EXCITE YOU ABOUT THE FIELD

“I have been in the field of artificial intelligence for about years, and it’s been truly rewarding. So I want you to participate in the beauty and the excitement of AI so you can become a professional who gets the same reward and excitement out of this field as I do.” Peter Norvig

STRUCTURE OF THIS COURSE:
*VIDEOS
*QUIZZES (which will ask you about your ability to answer AI questions)
*ANSWER VIDEOS ( which we tell you what the right answer would have been for the quiz that you might have falsely or incorrectly answered before.)
*ASSIGNMENT (you get a homework assignment also in the form of quizzes)
*VIDEO EXAMS

QUIZ:
AI PROGRAM IS CALLED ?
a. WETWARE
b. FORMULA
c. INTELLIGENT AGENT

QUIZ: (answer)
AI PROGRAM IS CALLED
a. WETWARE
b. FORMULA
c. INTELLIGENT AGENT (CORRECT)

Intelligent agent gets to interact with an environment. The agent can perceive the state of the environment through its sensors, and it can affect its state through its actuators. The big question of artificial intelligence is the function that maps sensors to actuators. That is called the control policy for the agent. So all of this class will deal with how does an agent make decisions that it can carry out with its actuators based on past sensor data. Those decisions take place many, many times, and the loop of environment feedback to sensors, agent decision, actuator interaction with the environment and so on is called perception action cycle.

QUIZ:
AI HAS SUCCESSFULLY BEEN USED IN ?
a. FINANCE
b. ROBOTICS
c. GAMES
d. MEDICINE
e. THE WEB
f.  NONE OF THEM

 

QUIZ: (answer)
AI HAS SUCCESSFULLY BEEN USED IN
a. FINANCE (correct)
b. ROBOTICS  (correct)
c. GAMES  (correct)
d. MEDICINE  (correct)
e. THE WEB  (correct)
f.  NONE OF THEM

“Every time you try to write a piece of software, that makes your computer software smart, likely you will need artificial intelligence. In this course, Peter and I will teach you many of the basic tricks of the trade to make your software really smart.” Sebastian Thrun

Some basic terminology that is commonly used in artificial intelligence to distinguish different types of problems.

FULLY VERSUS PARTIALLY OBSERVABLE:

An environment is called fully observable if what your agent can sense at any point in time is completely sufficient to make the optimal decision. For example, in many card games, when all the cards are on the table, the momentary site of all those cards is really sufficient to make the optimal choice.

That is in contrast to some other environments where you need memory on the side of the agent to make the best possible decision. For example, in the game of poker, the cards aren’t openly on the table, and memorizing past moves will help you make a better decision.

To fully understand the difference, consider the interaction of an agent with the environment to its sensors and its actuators, and this interaction takes place over many cycles, often called the perception-action cycle.

For many environments, it’s convenient to assume that the environment has some sort of internal state. For example, in a card game where the cards are not openly on the table, the state might pertain to the cards in your hand. An environment is fully observable if the sensors can always see the entire state of the environment.

It’s partially observable if the sensors can only see a fraction of the state, yet memorizing past measurements gives us additional information of the state that is not readily observable right now.

So any game, for example, where past moves have information about what might be in a person’s hand, those games are partially observable, and they require different treatment. Very often agents that deal with partially observable environments need to acquire internal memory to understand what the state of the environment is.

DETERMINISTIC VERSUS STOCHASTIC:

Deterministic environment is one where your agent’s actions uniquely determine the outcome. So, for example, in chess, there’s really no randomness when you move a piece. The effect of moving a piece is completely predetermined, and no matter where I’m going to move the same piece, the outcome is the same. That we call deterministic.

Games with dice, for example, like backgammon, are stochastic. While you can still deterministically move your pieces, the outcome of an action also involves throwing of the dice, and you can’t predict those. There’s a certain amount of randomness involved for the outcome of dice, and therefore, we call this stochastic.

DISCRETE VERSUS CONTINUOUS:

A discrete environment is one where you have finitely many action choices, and finally many things you can sense. So, for example, in chess, again, there are finally many board positions, and finally many things you can do.

That is different from a continuous environment where the space of possible actions or things you could sense may be infinite. So, for example, if you throw darts, there’s infinitely many ways to angle the darts and to accelerate them.

BENIGN VERSUS ADVERSARIAL:

In benign environments, the environment might be random. It might be stochastic, but it has no objective on its own that would contradict the own objective.

So, for example, weather is benign. It might be random. It might affect the outcome of your actions. But it isn’t really out there to get you.

Contrast this with adversarial environments, such as many games, like chess, where your opponent is really out there to get you. It turns out it’s much harder to find good actions in adversarial environments where the opponent actively observes you and counteracts what you’re trying to achieve relative to benign environment, where the environment might merely be stochastic but isn’t really interested in making your life worse.

CHECKERS
FULLY OBSERVABLE/DETERMINISTIC/DISCRETE/ADVERSARIAL

POKER
PARTIALLY OBSERVABLE/STOCHASTIC/ADVERSARIAL

ROBOTIC CAR
PARTIALLY OBSERVABLE/STOCHASTIC/CONTINUOUS

Artificial Intelligence is the technique of uncertainty management in computer software. AI is the discipline that you apply when you want to know what to do when you don’t know what to do.

There’s many reasons why there might be uncertainty in a computer program. There could be a sensor limit. That is, your sensors are unable to tell me what exactly is the case outside the AI system. There could be adversaries who act in a way that makes it hard for you to understand what is the case. There could be stochastic environments. Every time you roll the dice in a dice game, the stochasticity of the dice will make it impossible for you to be absolutely certain of what’s the situation. There could be laziness. So, perhaps you can actually compute what the situation is, but you computer program is just too lazy to do it. Ignorance; many people are just ignorant of what’s going on. They could know it, but they just don’t care.  All of these things are cause for uncertainty.

AI is the discipline that deals with uncertainty and manages it in decision making.

One of the best successes of AI technology at Google has been the machine translation system.

In Unit 1;

Key Applications of Artificial Intelligence

Definition of an Intelligent Agent

4 Key Attributes for Artificial Intelligence (partial observability, stochasticity, continuous spaces, and adversarial natures)

Sources and management of uncertainty

Mathematical concept of rationality are touched any of these issues superficially but as this class goes on you’re going to dive into any of those and learn much more about what it takes to make a truly intelligent AI systems.

You May Also Like

Leave a Reply

Your email address will not be published. Required fields are marked *