You are on page 1of 6

Waleed Amir

02-235171-041
Artificial
Intelligence
a) Define the concept of a Fully Observable environment.
In a fully observable environment, the state of a system is known at all times. Chess is a
good example. If you look at a chess board, all information required to make an optimal
decision is available. It doesn’t matter what happened in the past.

b) Consider a game of chess. Is this a deterministic or stochastic


environment? Clearly explain why.

Chess is deterministic environment if the opponent is deterministic. If the opponent isn't


guaranteed to take the same move given a specific state (i.e., board configuration), then
the environment is no longer deterministic.

If taking an action results in a non-deterministic move (e.g., taking the action to move the
queen 3 spaces may end up moving it 2, 3, or 4 spaces, etc.), then the environment is
certainly stochastic.
But the state your agent receives after taking an action comes after your opponent makes
a move. So, for the environment to be deterministic, the opponent will have to always
respond to each of your agent's moves in exactly the same way, otherwise the next state is
scholastically determined.
Chess, itself, is a deterministic game, but, depending on how you define the environment,
may not be a deterministic MDP. If you can guarantee the opponent is deterministic, then
it would be deterministic. If you can't, though, then, really, the problem becomes a
Markov game, which is significantly more complex than a deterministic MDP.

Chess is deterministic environment if the opponent is deterministic. If the opponent isn't


guaranteed to take the same move given a specific state (i.e., board configuration), then
the environment is no longer deterministic. If taking an action results in a non-
deterministic move (e.g., taking the action to move the queen 3 spaces may end up
moving it 2, 3, or 4 spaces, etc.), then the environment is certainly stochastic.

But the state your agent receives after taking an action comes after your opponent makes
a move. So, for the environment to be deterministic, the opponent will have to always
respond to each of your agent's moves in exactly the same way, otherwise the next state is
scholastically determined.
Chess, itself, is a deterministic game, but, depending on how you define the environment,
may not be a deterministic MDP. If you can guarantee the opponent is deterministic, then
it would be deterministic. If you can't, though, then, really, the problem becomes a
Markov game, which is significantly more complex than a deterministic MDP.
c) How abstraction may help to represent a problem in AI? What is a good
abstraction? Explain with an example
Abstraction has proven to be a powerful tool for controlling the combinatorics of a
problem-solving search. Problem solving using abstract versions of tasks can provide
cost-effective search heuristics and evaluations for the original, or "full", tasks which
significantly reduce their computational complexity, and thus make large problems
tractable

Abstraction occurs in this framework in the service of control decisions. If an impasse


occurs because of a lack of knowledge about how to make a selection, the resulting
search is performed abstractly. Consider a simple example from a toy robot domain.
Suppose that among the operators in the domain are ones that allow a robot to push a box
to a box (push-box-to-box(BOX1,130X2)) and go to a box (goto (BOX)). The
preconditions of the push-box-to-box operator are that the robot is next to the first box
and that the boxes are in the same room. The precondition of the goto operator is that the
robot is in the same room as the box that it wants to reach. The goal is to reach a state
where the two boxes (box l and box2 ) are next to each other. In the initial state, the robot
and the two boxes are in the same room together, but at different locations. Given this
initial state, a control decision must be made that will result in the selection of an
operator. With partial means-ends control knowledge (encoded as rules), the system can
determine that the push-box-tob ox(boxl,box2 ) operator is one of the possible
alternatives, but it may not be able to eliminate all of the other alternatives, leading to an
impasse, and thus a sub goal. In the sub goal, a search will be performed by trying out
each alternative until one is found that leads to the goal. When the push-box-to-box
operator is tried, it will fail to apply because one of its preconditions — that the robot is
next to the first box — is not met. However, if this precondition is abstracted away, then
the operator can apply abstractly, as the robot itself couldn't actually do this and the goal
of the abstract search will be achieved. From this abstract search, the information that the
push-box-to-box operator is the right one to select is returned, and used to make the
original control decision. Simultaneously, a control rule is learned which summarizes the
lesson of the abstract search.

Figure blow, illustrates this process.


A) The initial state contains three rings on the leftmost pole. These rings
must be moved to the rightmost pole as illustrated in the goal state. At
most one ring can be moved at a time. A larger ring cannot be placed on
top of a smaller ring. Rings can be placed on all three poles.
There are 3 poles and rings we need to shift the rings according to the same manner from first pole
to the third pole
Steps

Place Number 1 into the third pole from 1st pole


Number 2 will be placed into the second pole
Place Number 1 ring from third pole on ring number 2 into second pole
Place Number 3 ring from the first pole into the third pole

Number 1 ring will be pick from second pole and place into the first pole

Number 2 ring will be pick from second pole and place into the third pole

Number 1 will be pick from first pole and place into the third pole [goal state]

B) Determine the state space that must be performed to get from the initial
state to the goal state.

Answer
State: larger ring cannot be place on smaller ring, from one pole ring can be shift
to the next two poles

Initial state: all three ring will be on the first pole.

Successor function: state that contain actions are pick and drop pole to pole

GOAL state: three ring all in third pole with the numbering of 1on top 2 on
second place and 3 on bottom.

Q3) What is the current state of the art in artificial intelligence?


There is no doubt that artificial intelligence, machine learning, and data science have become the
most powerful and forward looking force in technology over the past decade. These technologies
have allowed for breakthrough insights and applications that may truly change the world for the
better. This is, of course, thanks to the symbiosis of data collection, hardware innovation, and
driven researchers that have taken hold over the 2010s. This has led us to bestow computers mind
boggling abilities in everything from vision to natural language processing to audio understanding
to complex signal processing. To understand where we are headed it is important to understand
how these breakthroughs took shape and where we currently stand.

3. What can AI do today?


a) Defeat a human champion in the board game Go.
No, although the computer programs are already challenging the highest ranking (9 dan)
professional players.
b) Drive a car in urban environment.
Yes, for example Google Self-Driving Cars have been in real world testing for several years.
c) Pass the Turing test.
Yes, although the test is very subjective and depends on the judges.
d) Clear all email spam.
No. Best learning algorithms are able to cope with most spam but a 100% accuracy is very hard
to accomplish and maintain. It is also very subjective what is considered as spam mail.
e) Translation of a book.
No, computer programs are able to translate many languages to some extent but high quality
translation of a book or other major piece of writing is still outside their capacity.
f) Clean your house.
Yes, millions of vacuum cleaner robots have been sold world wide, although most of them are
limited to floor level.
g) Plan the logistics of 50000 moving entities at a time.
Yes, for instance USA military used such planning and logistics software during the Desert
Storm military operation in 1991.
h) Fly a plane from take off to landing without any human intervention.
Yes, several unmanned aerial vehicles (UAVs) like Global Hawk are able to operate without
human intervention.
i) Do a complete surgery.
No, software-controlled robots are used to assist human surgeons but are not yet able to carry out
the complex operation fully by themselves.
j) Understand human speech.
No, the full comprehension of human speech requires human level AI. We are not there yet.

You might also like