From: Exploiting contextual information for prosodic event detection using auto-context
Auto-context (%)
None
Preceding
Following
Both
Two-way (M=3)
79.3
80.4
80.1
81.2 (M=3 itr = 3)
Four-way (M=2)
74.2
75.3
75.0
75.7 (M=2 itr = 6)