Dear soul-mates,

I would like to comment some issues of the QLearning algorithm in the book.

(1)
First of all I think there is a mistake in the pseudo code: For the update of
Q[a,s], r' instead of r should be used. Otherwise the values of the final states
will never be taken into account.

(2)
If my point (1) is correct, then the static variable r is not needed in the
code.

(3)
Nsa should be initialized to 0 for all values.

(4)
max(a') should be randomly chosen if the are more than one maximum (between
them)

(5)
f(q,n) must return a value even when q is null, i.e. when the agent has no idea
of the value of Q(a', s')
In the book it works because f(q,n) returns 2 the first 5 iterations, regardless
of the value of Q(a', s') (explained in the page 774 of the International
Edition)

(6)
I have been playing with the QLearning algorithm (after the modification
described in (1)) and the simple MDP world example. I have checked it with two
different parameters sets. The one described in the book:
     - reward of non-terminal states: -0.02 [r]
     - applies the value 2 [rp] to actions done less than 5 [en] times
     - the learning rate [rl] is
         60 / (60 - 1) + iteration
       i.e. the parameter rl is here 60
     - the number of trials is 2000 [tn]
and the values:
     lr = 5 // learning rate
     en = 100 ; // exploration number
     rp = 2 ; // value of unknown states
     tn = 300 ; // number of trials
     r = -0.05

I have computed the q-values of the four actions in state 3,3 for each
iteration. With the first set (see attached graph1.png for the values of a
representative experiment), in most of the experiments, the values are wrong at
the end.
With the second set, the values are correct for all experiments I have performed
so far (see attached graph2.png for an example)

Regards,
--
Ivan F. Villanueva B.
A.I. library:   http://www.artificialidea.com
<<<          The European Patent Litigation Agreement (EPLA)          >>>
<<<            will bring Software patents by the backdoor            >>>
<<<  http://www.no-lobbyists-as-such.com/florian-mueller-blog/epla/   >>>
<<<                     http://wiki.ffii.de/EplaEn                    >>>

can you help me please to answer this question !!

How can Artificial Intelligence technology be used in decision
support ?

                                        Thank you

Dear All,

Who know the answer for following question. It might change nowaday.

Read Turing¡¦s original paper on AI (Turing, 1950). In this paper,
he discusses several potential objections to his proposed enterprise
and his test for intelligence.
a. Which objections still carry some weight?

b. Are his refutations valid?

c. Can you think of new objections arising from developments since
he wrote the paper?

In the paper, he predicts that, by the year 2000, a computer will
have a 30% chance of passing a five-minute Turing Test with an
unskilled interrogator.
d. What change do you think a computer would have today?

Thanks

e. In another 50 years?

Hi,

I've a question about the difference of the model-based agent(mba) and
the goal-based agent (gba) without regarding the difference of reflex
and goal.

I mean the mba has informations about the world evolves, if he does an
action, right? But with this knowledge he knows about the world after
his action? I don't understand that the mba has no knowledge what the
world would be if he take an action.

Can someone explain it to me?

Greets Andi

Hi,

In page 25, it is said that "AI becomes a science (1987--present)".

I understand that it doesn't make much sense to pinpoint the exact
time when this process started, but on the other hand, the authors
must choose 1987 for a reason, which I cannot find in the following
description. Can anyone help?

Thanks in advance!

Pei

I don't have the book with me right now, but I believe we took the (as
you say, quite arbitrary) turning point to be Pearl's book, which has
1988 as a publication date, but actually appeared in 1987.

-Peter

On 11/1/06, Wang, Pei <Dr.PeiWang@...> wrote:
> Hi,
>
> In page 25, it is said that "AI becomes a science (1987--present)".
>
> I understand that it doesn't make much sense to pinpoint the exact
> time when this process started, but on the other hand, the authors
> must choose 1987 for a reason, which I cannot find in the following
> description. Can anyone help?

Hi Peter.

It is so nice to have an expert at hand.  I am just trying to get on the AI wagon.  I was reading some comments given by Dr. Roger Schank (who by chance I met down here in Panama City, Panama, Central América last year). So while we chat over a cup of coffe he said that the AI field is stalled due to the actual complexity for achiving the "goals" it has established for itself (since the begining), and eventhough he believes the goals can be achivied, it would take a lot deal of resources to get it (which probably nobody is willing to do) --- I hope I "tell the story the way it is" ...

I would trully appreciate your opinion on the matter ... to get some guidence from another quite qualified expert on the field.

Thanks ...

----- Message from peter@... ---------

Explain Gradient Descent Search. Is it optimal? its advantages
/disadvantages.

Does admissibility imply monotonicity and vice versa?if not then plz
tell when monotonicity imply admissibility.

Give arguments with 10 references (books / web / research paper)
about one of the following statements.

"Agents are future of AI"
  OR
"Agents are not AI"

The group is NOT a place for getting homework done.

regds,
Ravi
--- In aima-talk@yahoogroups.com, "it_expert_hina"
<it_expert_hina@...> wrote:
>
> Give arguments with 10 references (books / web / research paper)
> about one of the following statements.
>
> "Agents are future of AI"
>  OR
> "Agents are not AI"
>

Hi, I'm having trouble understanding this question.  How can there be
other connectives other than the ones already mentioned?  Can someone
give me an example please?

Thanks.

--- In aima-talk@yahoogroups.com, "it_expert_hina"
<it_expert_hina@...> wrote:
>
> Explain Gradient Descent Search. Is it optimal? its advantages
> /disadvantages.
>
Gradient Descent Search
http://en.wikipedia.org/wiki/Gradient_descent

-- AI

Hi,

you might want to look at what "binary connective" actually means.
hint:- Every possible truth table you can create using two variables
in effect defines a "binary connective"; and there are more than 4
such tables though most of them are useless (the question asks you to
think about why).

e.g: NAND or even something made up like UFALSE (defined as the
results are always FALSE irrespective of the values of A AND B).

Hope that helps,
Regards,
  ravi
--- In aima-talk@yahoogroups.com, "Julie Avery" <cookiesncream85@...>
wrote:
>
> Hi, I'm having trouble understanding this question.  How can there be
> other connectives other than the ones already mentioned?  Can someone
> give me an example please?
>
> Thanks.
>

In Sec. 13.7, "The Wumpus World revisted", the approach starts with
the distribution of all Pij variables, but only three of the Bij
variables, B11, B12, and B21.  It later factors out the "other"
variables, meaning only the "other" Pij ones.  Since there are
dependencies involving the "other" Bij ones, why are they not
included?  For example, there is a certain relation between P44 and
B34 which is part of the full joint.  The "other" Bij ones would also
factor out, since the results of interest are conditionally
independent of them also.

It might have been better to say that "other" includes all "other"
Pij *and* Bij ones, would it not?  Even though pits are the *cause*
of breezes, that doesn't mean that the Bij ones can simply be ignored.

   --  Bob Futrelle

Hello,
I would like to introduce an alternative java code for most of the algorithms in
the book. The code is extensively documented, licensed under the GPL and with
applets that can be used for teaching and learning. A list can be found in:

http://www.artificialidea.com/index.php?page=algorithms

Comments and help to improve it are welcome.
--
Iván F. Villanueva B.
A.I. library:   http://www.artificialidea.com
<<<          The European Patent Litigation Agreement (EPLA)          >>>
<<<            will bring Software patents by the backdoor            >>>
<<<                     http://epla.ffii.org/                         >>>

A quick glance suggests that Iván's software could be very useful to all of us.

I look forward to examining it in more detail and
to the reactions that others on this list might
have.

   - Bob Futrelle
_______________________________________________________________
Robert P. Futrelle      | Biological Knowledge Laboratory
     Associate Professor  | College of Computer  and Information
                          |     Science MS WVH202
Office: (617)-373-4239  | Northeastern University
Fax:    (617)-373-5121  | 360 Huntington Ave.
futrelle@...    | Boston, MA 02115
           http://www.ccs.neu.edu/home/futrelle
    http://www.bionlp.org   http://www.diagrams.org
_______________________________________________________________

>Hello,
>I would like to introduce an alternative java
>code for most of the algorithms in
>the book. The code is extensively documented, licensed under the GPL and with
>applets that can be used for teaching and learning. A list can be found in:
>
><http://www.artificialidea.com/index.php?page=algorithms>http://www.artificiali\
dea.com/index.php?page=algorithms
>
>Comments and help to improve it are welcome.
>--
>Iván F. Villanueva B.
>A.I. library: <http://www.artificialidea.com>http://www.artificialidea.com
><<< The European Patent Litigation Agreement (EPLA) >>>
><<< will bring Software patents by the backdoor >>>
><<< <http://epla.ffii.org/>http://epla.ffii.org/ >>>

--

I have a very fundamental question.

Figure 2.3 has the following percept sequence:

[A, Clean] [A, Clean] [A, Clean]

I feel that this percept sequence is not possible because after the
first [A, Clean] percept the agent will be in [B, xx] therefore it
will not receive the [A, Clean] percept.

Any help regarding this fundamental problem will be helpful.
Meanwhile, I will read the material more carefully once more.

Sincerely,

Lalit Gidwnai

my final project of AI course is about AI in chess,i searched and
found many resources but isn't so good :(

anyone can help me ?

1. AI Chess Project.
2. Chess Artificial Intelligence.
3. http://en.wikipedia.org/wiki/List_of_chess_engines

Enjoy

its general can u specific some points ??

 

 

On 1/7/07, v_r_boy <hadi.farnoud@... > wrote:

my final project of AI course is about AI in chess,i searched and
found many resources but isn't so good :(

anyone can help me ?

--
Eng. Osama Lotf Barakat

Technical Administrator;
www.ebnalyaman.com;

4th year ;
computer engineering;
Cairo Universty

On Mon, Jan 08, 2007 07:06:40PM +0330, hadi farnoud wrote:
> thanks ahmed,usefull links :)
> my project is research about AI chess algorithms.

Let us see the results. Ich you like an open source simplest chess engine for a
funny chess variant, you can use my applet and download the sources at:

Applet:
http://www.artificialidea.com/software/chess4/

Sources:
http://www.artificialidea.com/software/ai-20060928-beta.tgz

Explanation of the rules:
http://www.artificialidea.com/author/wiki/doku.php?id=quadchess
--
Iván F. Villanueva B.
A.I. library:   http://www.artificialidea.com
<<<          The European Patent Litigation Agreement (EPLA)          >>>
<<<            will bring Software patents by the backdoor            >>>
<<<                     http://epla.ffii.org/                         >>>

i found this:http://www.cs.biu.ac.il/~davoudo/
2 good research.

--- In aima-talk@yahoogroups.com, "Ivan F. Villanueva B." <ivan@...>
wrote:
>
> On Mon, Jan 08, 2007 07:06:40PM +0330, hadi farnoud wrote:
> > thanks ahmed,usefull links :)
> > my project is research about AI chess algorithms.
>
> Let us see the results. Ich you like an open source simplest chess
engine for a
> funny chess variant, you can use my applet and download the sources at:
>
> Applet:
> http://www.artificialidea.com/software/chess4/
>
> Sources:
> http://www.artificialidea.com/software/ai-20060928-beta.tgz
>
> Explanation of the rules:
> http://www.artificialidea.com/author/wiki/doku.php?id=quadchess
> --
> Iván F. Villanueva B.
> A.I. library:   http://www.artificialidea.com
> <<<          The European Patent Litigation Agreement (EPLA)
   >>>
> <<<            will bring Software patents by the backdoor
   >>>
> <<<                     http://epla.ffii.org/
   >>>
>

I have problem with implementation of uninformed search algorithms in
java that was in the aima.zip package.
Can anyone send for me a program that use this package to solve a
problem.
Thanks a lot.
Dellnya@...

Hi,

On page 74 of The Book there is a formula O(b^d+1).

The example states that b=10 and d=2, which in my mind results in 10^3
= 1000, but the example gives the result 1100.

Where is my mistake?

Regards,
Evyn

Hi,
The demo package in the distribution has examples of most if not all
the searches. The unit tests have more examples.

It is difficult to respond meaningfully to   reports of the type "I
have (some unspecified) problem". I am the maintainer of the java code
and if you would write to me directly (my email id is magesmail AT
yahoo DOT com . )explaining the precise problem you face, I might be
able to help.

Regards,
Ravi


--- In aima-talk@yahoogroups.com, "dellnya" <arman_fatahy@...> wrote:
>
> I have problem with implementation of uninformed search algorithms in
> java that was in the aima.zip package.
> Can anyone send for me a program that use this package to solve a
> problem.
> Thanks a lot.
> Dellnya@...
>