Skip to main content
. 2019 Aug 9;9(8):543. doi: 10.3390/ani9080543

Figure 5.

Figure 5

The DAG-HMM addressing the three-class problem of classifying cat manifestations. At each level, the remaining classes for testing are mentioned beside each node. Digging inside each node, an HMM-based sound classifier is responsible for activating the path of the maximum log-likelihood.