This preview shows page 1. Sign up to view the full content.
Unformatted text preview: on in ’doc.pdf’
Read new k -best ﬁle that is being emailed out.
NEW: Read the section in ’doc.pdf’ on the island algorithm. Prof. Jeﬀ Bilmes EE596A/Winter 2013/DGMs – Lecture 9 - Feb 6th, 2013 page 9-4 (of 180) Logistics Review On Final Project
Will be held Monday, March 18th, 2013.
Project should ideally be on some aspect of the material we have
learnt, some aspect of dynamic graphical models. Possible good
an implementation (i.e., a fast implementation of some DGMs
algorithm) and reporting and experience that you gain in doing this.
Application to real data.
A paper summary, of papers that we are not going to cover in this
A new idea of your own, new algorithms and/or theoretical results.
(e.g., approximation error for a sequential model).
Application of a DGM to a data domain (e.g., application of dynamic
Bayesian networks to speech/language/biology/surgery or some other
sequential data domain).
Applied use of GMTK on new applications domain.
Prof. Jeﬀ Bilmes EE596A/Winter 2013/DGMs – Lecture 9 - Feb 6th, 2013 page 9-5 (of 180) Logistics Review On Final Project The ideal project should be research-oriented, it is not acceptable to
propose whatever machine learning task you are c...
View Full Document
This document was uploaded on 04/05/2014.
- Winter '14