2.2 Decision stumps. A class that is often used to get a weak learner in boosting are Decision Stumps. We will see below that for this class an ERM rule is efficiently implementable, which additionally makes this class computationally attractive. Dec
Sorrow prepares you for joy. It violently sweeps everything out of your house, so that new joy can find
Idea Transcript
Institute of Software Engineering and Theoretical Computer Science
Machine Learning
Seminar Lerning Theory The seminar "Learning Theory" is an optional compulsory course (3 LP) in the module "Machine Learning II". Teaching coordination and course inscription are done via the ISIS site of the seminar. Detailed information on course formalities and accreditability are also summarized there. A description for external participants (Nebenhörer and Gasthörer) can be found below in the section on the online part of the course. Preliminary discussion and topic assignment
Friday, 12.04.2013, 14:00 am room: MAR 4.033
Seminar
the time schedule will be fixed in the discussion room: MAR 4.033
Responsible
Prof. Dr. Klaus-Robert Müller office hours: by appointment
Person of contact
Dr. Franz Király office hours: by appointment
Topics In the seminar, introductory topics in learning theory will be presented by the participants. Talks can be given in any of the following subject areas: Subject areas Vapnik-Chervonenkis Theory Statistical Learning Theory Learning with Neural Networks Compressed Sensing Learning with Algebraic Structure For single talk topics, the organizatorial schedule and supervisor assignments please consult the ISIS webpage of the seminar.
Prerequisites The following are optional prerequisites which are helpful but not necessary for participating in the seminar: Basic knowledge in machine learning, as presented in the respective modules (machine learning I or machine intelligence I) Basic knowledge in linear algebra and calculus, as presented in the respective modules (German: Lineare Algebra, Analysis) Basic knowledge in probability theory, as presentid in the module stochastics (German: Elementare Stochastik)
Accreditation The seminar "Learning Theory" can be accredited as an optional compulsory course for the module "Machine Learning II" and can be used to obtain part of the admission criteria for the exam. A successful participation in the seminar includes independently acquiring knowledge on the assigned topic, and preparing a 35 to 45 minute talk with slides on the topic (usually the slides are digital). For obtaining a certificate of participation, enrolment at the TU is neccessary in the form of a full enrolment, a Nebenhörerschaft or a Gasthörerschaft. For obtaining full accreditation, enrolment at the TU is neccessary in the form of a full enrolment or a Nebenhörerschaft. More details on registration, exam modalities and accreditation can be found on the online part of the course.
Online course / ISIS Course organization and participation in the online course are carried out over the ISIS system of the TU. For registration, a tubIT account is necessary. External participants who are enrolled at the TU (Nebenhörer and Gasthörer) can obtain an ISIS account at the tubIT office. For this, it is necessary to apply for an account at the tubIT-Laden with the Nebenhörer/Gasthörer enrolment certificate. The discussion forums which are visible to all participants and the anonymous feedback possibilities can be used by all participants of the online part for asking questions and providing comments.
Kontakt, Index und weiterer Service Last Update: 06.05.13
Kontakt, Inhaltsverzeichnis und weitere Service-Links