Re: For Patrick: Hons classes on MML from week 7

[Adam A. Ford]

Hi there,
I thought I might pass this info onto you about the MML class that David Dowe is teaching at Monash.
It was intended for Patrick but perhaps others might find it interesting as well.

On Tue, Sep 4, 2012 at 6:30 AM, David Dowe <david...@monash.edu> wrote:

Hi, Adam.

Please forgive the short notice, but I thought it might
possibly be of interest to you and/or to Patrick, a (3rd year)
CS undergrad' at the Univ' of Melbourne.   I think his name
is Patrick Robotham.  From his comments and question
on Sat 18/Aug/2012 at the Singularity Summit Australia
2012, it's clear that he has an interest in (algorithmic)
information theory  -  and (like the rest of us) still a few
things to learn.

Would you please be able to relay this e-mail on to him
(and apologies to both of you for the short notice)?

Starting in week 7 (Tue 4/Sept) going through to week 12,
I will teach the MML part of (the 4th Year Hons subject)
FIT4009 in Bldg 63, Room 126 (corner windowed room next
to 1st floor of stairwell) at Monash University (Clayton) on
Tuesdays from 2-4.   This will be a 12 hour
(6 weeks x 2hrs/week = 12 hours) introduction to the
Bayesian information-theoretic Minimum Message Length
(MML) principle and some of its developments since the
seminal Wallace & Boulton (1968) .

A rough outline of what I'll present is given in the Unit Guide at
http://infotech.monash.edu/units/archive/2012/s2/fit4009.pdf .

I haven't prepared slides.   Rather, I plan to use the

whiteboard and/or (possibly) overhead projector to go
through the likes of Chris Wallace's (2005) book

www.csse.monash.edu.au/~dld/CSWallacePublications#MMLBook

(which you should be able to obtain in hard copy

or soft electronic copy from the Monash library)
and other works, such as (e.g.) Dowe (2011a),

 "MML, hybrid Bayesian network graphical models, statistical consistency, invariance and uniqueness"

(which mentions, e.g., Comley & Dowe (2003, 2005), the first

two papers on Bayesian network graphical models which use

MML and which also have both discrete/categorical and
continuous-valued attributes).

No problems if you can't come (and likewise if you come
to some but not all), especially given the short notice.

If you are able to come and you are able to find the
time, please feel invited and most welcome to attend.

Again, no problems if you can't come (and likewise

if you come to some but not all), especially given the
short notice.

If you will attend or are thinking of attending, feel
welcome to have a look at the schedule in the
Unit Guide, and (although I make no promises)
feel welcome to make requests or suggestions
about what I present and/or how I present it.

As above, I haven't prepared slides but rather
plan to use the whiteboard and readings.

[For further evidence that Chris Wallace was a very+
smart+ guy, see, e.g.,
Christopher Stewart WALLACE (1933-2004) memorial special issue
[and front cover and back cover], Computer Journal
(Oxford U Press), Vol. 51, No. 5 (Sept. 2008),
with the lead article being my Dowe (2008a),
"Foreword re C. S. Wallace", pp523-560
(and here). www.doi.org: 10.1093/comjnl/bxm117.]

Thanks for your (possible) interest   and   best,

David.