Sorry, an error occurred while loading the content.

## many ambiguous questions in AIMA

Expand Messages
• for some reason, i really feel that some problmes in the textbook are really ambiguous, or hard to understand what it asks for. Or just because I am too
Message 1 of 7 , Nov 25, 2003
for some reason, i really feel that some problmes in the textbook
are really ambiguous, or hard to understand what it asks for. Or
just because I am too stupid.
For example in 18.8 (b)
show class probability p/(p+n) minimize the sum of squared errors.

what does class probability means?
does it mean the probability of the positive example occur in the
set?
in order to minimize a function, it must be that some varible takes
some value to minimize it. Here, who takes the value of p/(p+n)?

any reason why the question is kept hard to understand?

there are other examples in the textbook which i just could not
understand what exactly it asks for?

maybe i just should not learn AI?
• Could any one please explian to me the wupus example in chapter 6. -bhanu ... Do you Yahoo!? Free Pop-Up Blocker - Get it now
Message 2 of 7 , Nov 26, 2003

Could any one please explian to me the wupus example in chapter 6.

-bhanu

Do you Yahoo!?
Free Pop-Up Blocker - Get it now
• ... In chapter 6, the wupus example is a little difficult understand for the read of first time. It just gives the abstruct structure for how to describe and
Message 3 of 7 , Nov 27, 2003
--- In aima-talk@yahoogroups.com, Maithreebhanu <bhanu128@y...> wrote:
>
> Could any one please explian to me the wupus example in chapter 6.
>
> -bhanu
>
>
>
>
> ---------------------------------
> Do you Yahoo!?
> Free Pop-Up Blocker - Get it now

In chapter 6, the wupus example is a little difficult understand for
the read of first time. It just gives the abstruct structure for how
to describe and solve problem. I think because it is difficult to
implement this method, therefore the author doesn't give the exercise
for programming. But wupus example has been used in chapter 8 (if I
remember) and much better method has been used for handling this
problem.
Maybe you could continue read for comparing these chapters for
further understanding.
In addition, Mr. norvig has provided source code for "wumpus"
problems by lisp in the "agent" and "logic" (If I remember) folders,
you could check it. The java source code is also available for
this "wumpus" problem.

kind regards/chenyu
• thank you chenyu, I shall certainly follow your advice. But i have another question. In figure 2.14 (pg 48 ed1) the books gives psuedo code for the
Message 4 of 7 , Nov 29, 2003
thank you chenyu, I shall certainly follow your advice. But i have
another question. In figure 2.14 (pg 48 ed1) the books gives psuedo
code for the environment, the line

Action[agent]<-- Program[agent](Percept[agent])

is unclear. And how this maps to wupus world enviornment?
• This line of pseudocode says to take the agent s program and apply it to the agent s percept, in order to get the agent s action. If there were only one
Message 5 of 7 , Nov 29, 2003
This line of pseudocode says to take the agent's program and apply it
to the agent's percept, in order to get the agent's action. If there
were only one agent, it would be

action <-- program(percept())

It is connected to the environment in that the percept function is
determined by the environment.

On Saturday, November 29, 2003, at 07:42 AM, Maithreebhanu
Wimalasekare wrote:

> thank you chenyu, I shall certainly follow your advice. But i have
> another question. In figure 2.14 (pg 48 ed1) the books gives psuedo
> code for the environment, the line
>
> Action[agent]<-- Program[agent](Percept[agent])
>
• Thank you for helping me out. I find BNF and Psedo code used in this book to diificult to grasp at first. Are there any documents on them for us to learn
Message 6 of 7 , Dec 1, 2003
Thank you for helping me out. I find BNF and Psedo code used in this
book to diificult to grasp at first. Are there any documents on them
for us to learn these?

--- In aima-talk@yahoogroups.com, Peter Norvig <peter@n...> wrote:
> This line of pseudocode says to take the agent's program and apply
it
> to the agent's percept, in order to get the agent's action. If
there
> were only one agent, it would be
>
> action <-- program(percept())
>
> It is connected to the environment in that the percept function is
> determined by the environment.
• ... this ... them ... The book s writing style may be like this: a simple problem ( provide simple solution - a complex problem (provided modified solution or
Message 7 of 7 , Dec 1, 2003
--- In aima-talk@yahoogroups.com, "Maithreebhanu Wimalasekare"
<bhanu128@y...> wrote:
>
> Thank you for helping me out. I find BNF and Psedo code used in
this
> book to diificult to grasp at first. Are there any documents on
them
> for us to learn these?
>

The book's writing style may be like this: a simple problem ( provide
simple solution -> a complex problem (provided modified solution or
renewed solutions )-> more complex problem (provided better solution
or renewed solutions) - > more practical problem for thinking

In the above solutions, the author give the natural language
(English) to show the main idea for how to solve the problem. The
description is long but clear and easy to understand, but it is also
difficult to remember and for further analysis and implementation (by
lisp or python etc computer language).
Therefore, almost for the above every solutions, the author provides
Psedo code together, it is much shorter. Because it doesn't contains
description, if you only read the Psedo code for understanding, it is
difficult.

In addition, BNF doesn't appear (if I remember) until the "natural
langauge" related chapters. Therefore BNF will not become your
obstacle in learning (from "agent" chapter to "planning" chapter).

My first suggestion is to read the solution's description and Psedo
code together, not seperately, even if they are on the different
pages ( You need to turn pages repeatedly. ).

My second suggestion is to read patiently. It is impossible to learn
AI in 2 or 3 months, even for a genius. But it is worth to spend time.

kind regards/chenyu
>
>
>
>
>
>
>
> --- In aima-talk@yahoogroups.com, Peter Norvig <peter@n...> wrote:
> > This line of pseudocode says to take the agent's program and
apply
> it
> > to the agent's percept, in order to get the agent's action. If
> there
> > were only one agent, it would be
> >
> > action <-- program(percept())
> >
> > It is connected to the environment in that the percept function
is
> > determined by the environment.
Your message has been successfully submitted and would be delivered to recipients shortly.