Multiview Active Learning for Scene Classification with High-Level Semantic-Based Hypothesis Generation

Table 1

Performance comparison of different theme models.

	N1	N2	N3	N4	N5	N6	N7	N8
(a) OT image dataset
LDA
SP-pLSA
CO-LDA
	S1	S2	S3	S4	S5	S6	S7	S8
(b) UIUC Sports image dataset
LDA
SP-pLSA
CO-LDA

Note. Bold values represent the best performance of the algorithms corresponding to each class.