Table 4 Unweighted accuracy (%) of different methods in different cases
From: Cross-corpus speech emotion recognition using subspace learning and domain adaption
Case | Subspace learning | Distribution adaptation | Feature selection | The proposed method | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
GFK | PCA | LDA | KSR | SA | MEDA | JDA | TCA | BDA | TJM | ||
N-B | 39.1% | 33.75% | 31.67% | 36.67% | 37.71% | 30.42% | 33.33% | 32.71% | 36.86% | 38.75% | 48.54% |
B-N | 35.28% | 38.75% | 31.46% | 40% | 44.58% | 31.25% | 44.79% | 43.75% | 42.08% | 45% | 65.28% |
N-I | 43.75% | 41.86% | 32.71% | 27.92% | 37.5% | 38.96% | 45.42% | 50.42% | 31.67% | 30.63% | 53.33% |
I-N | 31.04% | 30.63% | 34.38% | 36.67% | 42.29% | 31.25% | 38.33% | 35% | 39.58% | 43.88% | 64.04% |
B-I | 42.79% | 38.54% | 45% | 48.96% | 38.33% | 39.17% | 34.42% | 45.42% | 37.92% | 42.92% | 52.28% |
I-B | 32.91% | 36.08% | 40.86% | 34.79% | 34.58% | 39.58% | 36.67% | 34.79% | 46.25% | 45.83% | 46.08% |
Average | 37.47% | 36.6% | 36.01% | 37.5% | 39.17% | 35.11% | 39.33% | 40.35% | 39.06% | 41.17% | 54.93% |