Research Article
Adaptive Online Sequential ELM for Concept Drift Tackling
Table 2
Data set dimension, quantity, evaluation method, and performance measurement.
| (a) Data set dimension and quantity |
| | Data set | Concepts | Inputs | Outputs | Quantity (×concepts) |
| | SEA | 4 | 3 | 2 | 20000 (×4) | | STAGGER | 3 | 9 | 2 | 4400 (×3) | | MNIST | 2 | 784, 865 | 10 | 70000 (×2) | | USPS | 1 | 865 | 36 | 48908 (×1) |
|
|
| (b) Evaluation method |
| | Data set | Evaluation method | Training | Testing |
| | SEA | 5-fold cross-validation | 16000 (×4) | 4000 (×4) | | STAGGER | 5-fold cross-validation | 3520 (×3) | 880 (×3) | | MNIST | Holdout (10x trials) | 60000 (×2) | 10000 (×2) | | USPS | Holdout (10x trials) | 35050 | 13858 |
|
|
| (c) Performance measurements |
| | Measure | Specification |
| | Accuracy | The accuracy of classification in % from | | Predictive accuracy | The accuracy measurement of the future sequential training data [20] | | Testing accuracy | The accuracy measurement of the testing data set excluded from the training | | Forgetting capability | The testing accuracy differences between the current concept with the previous concepts | | Cohen’s kappa and kappa error | The statistic measurement of interrater agreement for categorical items |
|
|
