Table 3 Weighted 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 | 35.63% | 32.08% | 30.63% | 37.29% | 36.46% | 35.00% | 36.88% | 31.04% | 37.29% | 37.50% | 50.42% |
B-N | 32.71% | 37.29% | 34.79% | 40.42% | 42.92% | 38.75% | 49.58% | 43.04% | 51.04% | 45.63% | 67.08% |
N-I | 43.13% | 40.83% | 47.08% | 47.92% | 37.29% | 42.29% | 43.54% | 57.50% | 44.17% | 46.04% | 61.25% |
I-N | 33.33% | 38.33% | 32.29% | 37.71% | 41.88% | 41.86% | 44.79% | 41.04% | 44.58% | 45.00% | 67.75% |
B-I | 41.46% | 39.58% | 46.67% | 50.21% | 37.71% | 47.71% | 33.96% | 43.33% | 38.54% | 46.04% | 55.83% |
I-B | 31.88% | 38.75% | 42.08% | 39.58% | 41.04% | 41.25% | 38.96% | 36.67% | 47.29% | 49.17% | 46.88% |
Average | 36.35% | 37.81% | 38.92% | 42.18% | 39.55% | 41.14% | 41.29% | 42.10% | 43.82% | 44.90% | 58.20% |