How hard was the exam?

Please post constructive comments here....

Past, Present and Future? of AI

I just read this article (http://www.techradar.com/news/world-of-tech/future-tech/the-past-present-and-future-of-ai-643838) about AI. It's a good read, but it focuses on the popular idea of AI from its humble beginnings. Nothing in here about planning, preference representation, or relational Bayes' nets. The article does at least acknowledge the disconnect between "Hollywood" AI and real AI research--something lacking in most mainstream coverage of AI.

Lisp type philosophy catching on fast

I have noticed this new trend where the advantages of pure functional programming are filtering into newer languages in the main stream. One specific pivot point is concurrency and scalability. The paradigm of how immutability==high concurrency==scalability is being replicated in multiple ways, first map reduce now this. Its high time the goodness of lisp like languages was not restricted due to its esoteric coding style ...
http://blog.bestinclass.dk/index.php/2009/09/scala-vs-clojure-lets-get-down-to-business/

Human-Aware AI: Robots that "show emotion"

This great talk on TED by David Hanson talks about robots and robot faces that interface with humans and are able to show, among other things, empathy.

http://www.ted.com/talks/david_hanson_robots_that_relate_to_you.html

Stanford's Autonomous Helicopter Project

Here's a link to Stanford's autonomous helicopter project, part of which Andrew Ng showcased at IJCAI 2009 during his Computers & Thought Award lecture.

http://heli.stanford.edu/

Remember, this helicopter is completely autonomous :)

Google code jam, easy AI problem!!?

This is a good problem in google code jam... Relates to AI kind of...
1)You know your initial state completely
2) Environment observability is only local(next rooom), but it is static
3) Actions are deterministic...
http://code.google.com/codejam/contest/dashboard?c=32003#s=p1

The problem boils down to this:
"given your actions can you determine the configuration of your environment, in this case a maze..."
Here is the python solution... Kinda had fun time solving it.. took about 1.5 hours to figure the problem and solve it... Have fun!

Next on the reading list (for classes next week)

Once we are done with Bryce el al paper on belief-space planning, we will turn to atomic models for
decision-theoretic search--aka Markov Decision Processes.

The primary reading will be the following paper:

http://www.cs.washington.edu/research/jair/abstracts/boutilier99a.html

You should read until the end of Section 3 for now (section 4 is factored representations--we will discuss
that later)

I will also provide you with the relevant chapter from R&N.

Rao

A short writeup on use of binary decision diagrams (BDDs) in planning

You may want to look at the short write at the following URL to get an idea of how bdds get used in planning.


http://rakaposhi.eas.asu.edu/cse571/bdd-intro.pdf

rao

Setup a wiki page for listing your presentation topic choices

I set up a wikipage accessible from the main 571 wikipage where you can write in your topic choice for reading/presentation.

It is also directly accessible via
 http://rakaposhi.eas.asu.edu/mediawiki/index.php/Page_for_listing_your_top_two_presentation_topics_from_IJCAI_proceedings

Rao

Tomorrow's class agenda (and the possibility of a make-up class on 9/11)

Folks:

 Tomorrow, we will complete the discussion of section 4.4 (belief-space search), and also cover 4.5 (online search).
If you haven't read section 4.5, please make sure you do.  Time permitting, we will start with Bryce paper. We will come back to it next week.

 We will also try to see if we can schedule a make-up class next week (perhaps on Friday the 11th). This is because I will be missing two classes on the week of 21st as I will be attending ICAPS 2009 in Greece (I know--someone has to do all this hard travel..), and prefer to make them up myself.  So, please come with your Friday 11th calendar.

Rao

New comment on Stochastic vs. Limited Senses.

The following comment sent last semester to 471 folks is relevant given the short philosophical discusson yesterday on this topic.

rao

ps: The full thread is available at  http://cse471-s09.blogspot.com/2009/01/stochastic-vs-limited-senses.html

---------- Forwarded message ----------
From: Subbarao Kambhampati <noreply-comment@blogger.com>
Date: Tue, Jan 27, 2009 at 7:42 PM
Subject: [cse471/598 Intro to AI Spring 2009 Blog] New comment on Stochastic vs. Limited Senses.

Jason Majors asks:
 
 =============

Tuesday, January 27, 2009

Stochastic vs. Limited Senses

While reading through the book today, I came up with a question. Where's the line between a lack of good senses and a stochastic environment? The examples we've had in class of what makes an environment stochastic (e.g. car trouble with the taxi) could be monitored with an appropriately advanced sensor array. An omniscient agent (or omniscient objective observer) would consider every environment fully deterministic.
So is there a clear line between the two? I would think that something that is beyond practical (such as monitoring the surfaces of a taxi's tires to know when it will weaken enough to burst from the pressure) would be in the stochastic column, but what about not knowing that a tire is going flat, because the taxi lacks a tire pressure monitoring system (which is practical)?

==============


Subbarao Kambhampati has left a new comment on your post "Stochastic vs. Limited Senses":

This is a good question. When we do not have complete models of a world, then what is inherently a "deterministic world" may well look like a non-deterministic/stochastic one to the agent.

About the only "natural" world that can be said to be inherently stochastic is again the quantum world. In every other case, you can--if you prefer--think that the underlying world is deterministic and we just didn't model it adequately. [Even in the quantum case, many scientists--Einstein in particular--fought tooth and nail to convince the scientific community that the uncertainty is *not* inherent and that we get it only because we are not modeling the system completely. See the celebrated EPR paradox--and how it was eventually shown that the "paradox" is really not a paradox and quantum uncertainty is very much inherent: http://en.wikipedia.org/wiki/EPR_paradox ]


Notice that the "completeness" we are talking about doesn't have to involve as high a granularity as "modeling the imperfections on the coin surface and the eddies in the room air to predict the coin toss outcome".

Remember that our ancestors, not too long ago, assumed that many phenomena that we now know as deterministic--such as eclipses-- are actually non-deterministic (and thus would associate with them superstitions like "the eclipse shows the gods being angry with the ruler"--as in Chinese belief of "Mandate from the Heaven" etc).



Rao

ps: Of course, we humans are also equally adept at introducing determinism where there is none (e.g. my favorite god created this whole darned entire universe and all its life forms over a weekend some 6000 years ago....

Check out http://www.tv.com/the-simpsons/lisa-the-skeptic/episode/1471/summary.html

where Lisa laments that the expected answer to every question on her science test was "God made it" ;-)



Posted by Subbarao Kambhampati to cse471/598 Intro to AI Spring 2009 Blog at January 27, 2009 6:42 PM

Re: Wikinotes: An experimental feature for the class

On second thoughts, I think it will be useful to have a way of knowing who made what changes to the wikinotes (if nothing else, this can be particpation credit!).

So, you can create yourself an account using the following link

http://rakaposhi.eas.asu.edu/mediawiki/index.php?title=Special:UserLogin&returnto=Special:UserLogin

Then when you edit, login before editing.

Let's see how web3.0 you can get.

thanks
Rao

On Mon, Aug 31, 2009 at 7:56 PM, Subbarao Kambhampati <rao@asu.edu> wrote:

Folks:

 I installed a mediawiki page for the class so you can collaboratively develop wikinotes for the class. There is a link to the wikinotes from the class page, but it is also available at
http://rakaposhi.eas.asu.edu/mediawiki/index.php/Cse571

Notice that currently, like wikipedia, anyone can edit pages. If we have software engineering and/or graphics students editing our notes, I will consider installing password permissions.

I assume you all are familiar with wikipedia page editing--mediawiki basically uses the same editor--very easy to use.

I setup a skeletal page for today's lecture and wrote a short summary of the observability issue in classical planning. Others can add anything they particularly remembered/understood.

cheers
Rao

Wikinotes: An experimental feature for the class

Folks:

 I installed a mediawiki page for the class so you can collaboratively develop wikinotes for the class. There is a link to the wikinotes from the class page, but it is also available at
http://rakaposhi.eas.asu.edu/mediawiki/index.php/Cse571

Notice that currently, like wikipedia, anyone can edit pages. If we have software engineering and/or graphics students editing our notes, I will consider installing password permissions.

I assume you all are familiar with wikipedia page editing--mediawiki basically uses the same editor--very easy to use.

I setup a skeletal page for today's lecture and wrote a short summary of the observability issue in classical planning. Others can add anything they particularly remembered/understood.

cheers
Rao

Next reading

Once we are done with "Beyond Classical Search" we will shift to belief-space planning. The paper for that is

http://rakaposhi.eas.asu.edu/dan-jair-pond.pdf

This is a long paper. You should certainly get started on sections 2 and 3.

Rao

Discussion question on Conformant and Contingent search (add your comments on the blog)

So you all have read the chapter "Beyond Classical Search" which talks about search strategies to use when we have non-deterministic actions,
and partial observability or unobservability.

You saw that the first two can be handled by And-Or search--first in the space of states while the second in the space of belief states. The third can be done
by normal A* search in the space of belief states.

The chapter is somewhat coy on what you will have to do regarding heuristics. Afterall, we spent a lot of time talking about heuristics for A* in the space of states. How about for these beasts? This is what I want you to think about.

To think of something concrete, consider the scenario of searching in a 2-D space (which possibly has obstacles). Now, if we consider deterministic actions and full observability in this space, we basically just have A* search. A good adminissible heuristic is "straight line-distance".

Now, suppose you are still in the same space but

1. Your actions are not deterministic (but you have full observability)

2. You have partial observability

3. You have no observations


In each of the scenarios above, explain how you can use and/or generalize the straightline distance heuristic.

Please write your thoughts as comments to this question on the blog.

thanks
Rao

so apparently Gail Collins at NYT got wind of all our discussion about Human Aware AI...

FYI
   http://www.nytimes.com/2009/08/29/opinion/29collins.html?em


rao

References from last class

The paper on "Unreasonable effectiveness of data" that I mentioned in the class is at

http://rakaposhi.eas.asu.edu/cse571/unreasonable.pdf

(the original Wigner paper on "unreasonable effectiveness of Mathematics in Physical Sciences" can be read at
http://www.dartmouth.edu/~matc/MathDrama/reading/Wigner.html )

The paper on "intelligence without representation"--by Brooks--written when Top Down approach was more in vogue--is at

http://people.csail.mit.edu/brooks/papers/representation.pdf

(a related paper titled "Intelligence without robots" is also interesting--but again not as surprising given the amount of AI efforts directed to the web thesedays:
http://www.aaai.org/ojs/index.php/aimagazine/article/viewFile/1065/983
)

Here is a short paper on computational models of emotions and how they can be used

https://www.aaai.org/Papers/AAAI/2006/AAAI06-263.pdf

(Here is a talk I gave sometime back on "human-aware AI"
 http://rakaposhi.eas.asu.edu/uw-haai-talk.ppt  (slides
 http://rakaposhi.eas.asu.edu/rao-uw-talk.WAV (audio)
)

The Seinfeld episode where uncle leo's eyebrows cause miscued emotional recognition in his audience is at

http://www.youtube.com/watch?v=FevMimZyeLI


Rao

Class blog setup...

Folks:
  I set up a blog for the class. All mails sent to the class will be archived on the blog (as well as on a mail archive available from the class page).   You have all been sent invitations to the blog.

    The first "discussion" question follows.

Rao