Title :
Classifier-based affinities for clustering sets of vectors
Author :
García-García, Darío ; Santos-Rodríguez, Raúl ; Parrado-Hernández, Emilio
Author_Institution :
Ind. Syst. Unit, TECNALIA Res. & Innovation, Donostia - San Sebastián, Spain
Abstract :
We focus on the task of clustering sets of vectors. This can be seen as a special case of sequence clustering when the dynamics are not taken into account. We propose to use the error probability of binary classifiers to obtain a measure of the affinity between two sets so that a standard similarity-based clustering algorithm can be applied.
Keywords :
error statistics; pattern classification; pattern clustering; vectors; binary classifiers; classifier-based affinities; error probability; similarity-based clustering algorithm; vector clustering sets; Hidden Markov models; Kernel; Measurement uncertainty; Probabilistic logic; Probability distribution; Standards; Vectors; Sequence clustering; sets of vectors; speaker clustering;
Conference_Titel :
Machine Learning for Signal Processing (MLSP), 2012 IEEE International Workshop on
Conference_Location :
Santander
Print_ISBN :
978-1-4673-1024-6
Electronic_ISBN :
1551-2541
DOI :
10.1109/MLSP.2012.6349760
