== Praktikum Maschinelles Lernen und Datenanalyse ==

 || '''Schedule:''' || Irregular, see  [[https://www.google.com/calendar/embed?src=pd5q3lv0b7833f3lnebbmqudjc%40group.calendar.google.com&ctz=Europe/Berlin|calendar]] ||
 || '''Room:'''   || FR 6046 (Mondays) und FR 6043 (Wednesdays)   ||
 || '''Lecturer''' || Prof. Dr. Klaus-Robert Müller ||
 || '''Contact:''' || [[http://www.user.tu-berlin.de/paulbuenau/|Paul von Bünau]] ||
 || '''Module:''' || M.Sc. Module Praktikum Maschinelles Lernen und Datenanalyse ||

The aim of this lab course is to practice the process of explorative data analysis and understand the main algorithms. 
The focus is on dimensionality reduction, clustering, classification and regression. For each assignment, a number
of algorithms have to be implemented (in Matlab) and analyzed in experiments on real or synthetic data. Taking the 
[[Main/SS11_MatlabKurs|Matlab course]] and the [[Main/WS10_MaschinellesLernen1|Machine Learning lecture]] is highly 
recommended but not a formal prerequisite. 

The lab course consists of two parts: a lecture (Mondays at 10.15am), which is held only when a new assignment is handed out, 
and a consultation on Wednesdays (10.15am) in all following weeks. Please have a look at the [[https://www.google.com/calendar/embed?src=pd5q3lv0b7833f3lnebbmqudjc%40group.calendar.google.com&ctz=Europe/Berlin|calendar]] for the exact schedule.

Please register in the [[http://groups.google.com/group/mikiobraun-lehre|google group]] to receive announcements 
and ask questions.

More information can be found in the german [[attachment:Praktikum_ML_Info.pdf|Informationsblatt.]]

=== Schedule ===

See [[https://www.google.com/calendar/embed?src=pd5q3lv0b7833f3lnebbmqudjc%40group.calendar.google.com&ctz=Europe/Berlin|calendar]]

=== Material ===

   * [[attachment:ML_Praktikum_U01_en.pdf|Problem sheet #1 (Matlab)]], tests: [[attachment:U01_test_distmat.m|U01_test_distmat.m]],[[attachment:U01_test_mydet.m|U01_test_mydet.m]], [[https://ml01.zrz.tu-berlin.de/~paul/pass.pl?conf=ss11_prak01.conf|PASS]]

   * [[attachment:ML_Praktikum_U02_en.pdf|Problem sheet #2 (unsupervised learning)]], Data: [[attachment:flatroll.mat|flatroll.mat]], [[attachment:fishbowl_swissroll_correct.mat|fishbowl_swissroll_correct.mat]], [[attachment:U01_usps.mat|U01_usps.mat]]. [[attachment:lle_talk.pdf|LLE slides]]. Tests: [[attachment:U02_tests.zip|U02_tests.zip]], [[https://ml01.zrz.tu-berlin.de/~paul/pass.pl?conf=ss11_prak02.conf|PASS]]

   * [[attachment:ML_Praktikum_U03.pdf|Problem sheet #3 (Clustering, EM)]], [[attachment:ML_Praktikum_U03_en.pdf|English version]], Data: [[attachment:U03_5gaussians.dat|U03_5gaussians.dat]], [[attachment:U03_2gaussians.dat|U03_2gaussians.dat]], [[https://ml01.zrz.tu-berlin.de/~paul/pass.pl?conf=ss11_prak03.conf|PASS]], Tests: [[attachment:U03_tests.zip|U03_tests.zip]]

  * [[attachment:ML_Praktikum_U04.pdf|Problem sheet #4 (Classification: KRR, CV, ROC)]], [[attachment:ML_Praktikum_U04_en.pdf|English version]], [[https://ml01.zrz.tu-berlin.de/~paul/pass.pl?conf=ss11_prak04.conf|PASS]], Tests:  [[attachment:U04_test_CV_krr.m|U04_test_CV_krr.m]], [[attachment:U04_datasets.tar.gz|Datasets]]

  * [[attachment:ML_Praktikum_U05_en.pdf|Problem sheet #5 (SVMs)]], [[attachment:U05_datasets.zip|Datasets]],  [[https://ml01.zrz.tu-berlin.de/~paul/pass.pl?conf=ss11_prak05.conf|PASS]]

==== Literature ====

[[attachment:guide.pdf|Lecture notes (as of April 2010)]]

==== Results ====

|| '''Matrikelnr.''' || '''Sheet 1''' || '''Sheet 2''' || '''Sheet 3''' || '''Sheet 4''' || '''Sheet 5''' ||
||305493||5||20||15||19||20||
||307336||5||20||18||18||20||
||306383||5||20||18||18||20||
||331086||5||22||20||18||20||
||310234||5||22||20||18||20||
||310260||5||17||19||20||15||
||311217||5||17||19||20||15||
||315728||5||18||20||20||20||
||314519||5||18||20||20||20||
||327898||5||20||20||15||18||
||329282||5||20||20||15||18||
||329499||5||20||20||20||20||
||329478||5||20||20||20||20||
||334450||5||21||21||15||20||
||335614||5||21||20||19||15||
||336016||5||21||20||19||15||

=== Access to Matlab ===

The servers {{{ {bolero,pepino,fiesta}.cs.tu-berlin.de }}} can be reached via ssh.