EURASIP Journal on Audio, Speech, and Music Processing

Table 2 PCA baseline classification accuracy (%)

From: High level feature extraction for the self-taught learning algorithm

Training	Dictionary size K
Set	100	200	300	500
IS-20	38.5	36.8	36.5	35.6
IS-50	43.5	41.8	41.8	41.7
IS-100	54.3	52.3	53.3	53.0
IS-250	56.0	56.3	56.6	57.3

Bases are learned from the labeled ISMIR training data.

Back to article page