TY - GEN
T1 - Characterizing time series variability and predictability from information geometry dynamics
AU - Dubnov, Shlomo
PY - 2013/10/8
Y1 - 2013/10/8
N2 - This paper presents a method for analyzing changes in information contents of time series based on a combined adaptive approximate similarity detection and temporal modeling using Bregman information. This work extends previous results on using information geometry for musical signals by suggesting a method for optimal model selection using Information Rate (IR) as a measure of an overall model predictability.
AB - This paper presents a method for analyzing changes in information contents of time series based on a combined adaptive approximate similarity detection and temporal modeling using Bregman information. This work extends previous results on using information geometry for musical signals by suggesting a method for optimal model selection using Information Rate (IR) as a measure of an overall model predictability.
UR - http://www.scopus.com/inward/record.url?scp=84884946515&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-40020-9_73
DO - 10.1007/978-3-642-40020-9_73
M3 - Conference contribution
AN - SCOPUS:84884946515
SN - 9783642400193
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 658
EP - 665
BT - Geometric Science of Information - First International Conference, GSI 2013, Proceedings
T2 - 1st International SEE Conference on Geometric Science of Information, GSI 2013
Y2 - 28 August 2013 through 30 August 2013
ER -