"Vectors of high dimensionality". Is this interesting to J ? See below.


-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of 
Tomi Kinnunen
Sent: 26. marraskuuta 2013 10:33
To: [email protected]
Subject: NIST i-vector machine learning challenge @ Odyssey 2014

Hi all,

Apologies for possible cross-postings.

Attention, machine learning people: NIST organizes a new type of machine 
learning challenge. The first results of the challenge are presented as a part 
of Odyssey 2014: The Speaker and Language Recognition Workshop conference that 
UEF hosts in 2014
(http://cs.uef.fi/odyssey2014/)

National Institute of Standards and Technologies (NIST) organizes a new type of 
"i-vector machine learning challenge":

http://www.nist.gov/itl/iad/mig/ivec.cfm

The purpose is to invite researchers outside of speech & audio processing 
fields to do voice biometrics (speaker recognition) without any knowledge 
required of speech or signal processing. It involves processing of feature 
vectors of high dimensionality, each being a descriptor of one long audio 
record. NIST shares an unlabeled training set and participants will then use an 
online platform to submit pairwise similarity scores of vectors that NIST 
scores.

Schedule:

Late November 2013: Registration opens on website Late November 2013: Challenge 
data available on website [ February 10, 2014: Odyssey papers on Challenge due 
-- OPTIONAL ] April 7, 2014: Last day to submit output for official scoring 
April 8, 2014: Official scores (on evaluation subset) posted

Feel free to forward this to your colleagues as widely as you can!

Best wishes,
Tomi

--
==========================================
Dr. Tomi H. Kinnunen, Researcher
School of Computing
University of Eastern Finland
P.O. Box 111, FI-80101 Joensuu, FINLAND
Email: [email protected], [email protected]
WWW: http://cs.joensuu.fi/pages/tkinnu/webpage/
Tel. +358 50 442 2647, +358 44 507 0624
==========================================

----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to