Research Article
Social Group Behavior Analysis Model Integrating Multitask Learning and Convolutional Neural Network
Table 5
Comparison of deep learning method and nondeep learning method on classified data sets.
| | Accuracy | Precision | Recall | F1 | TNR |
| Nondeep learning method | LapRLS | 0.635 | 0.534 | 0.548 | 0.667 | 0.917 | KSRC | 0.831 | 0.792 | 0.853 | 0.814 | 0.954 | mSRC | 0.877 | 0.844 | 0.923 | 0.872 | 0.963 | MCMI-AB | 0.618 | 0.595 | 0.574 | 0.564 | 0.898 | mcSVM | 0.684 | 0.698 | 0.578 | 0.577 | 0.922 | ESRC | 0.801 | 0.731 | 0.885 | 0.778 | 0.941 | Deep learning methods | CNN + SVM | 0.969 | 0.969 | 0.969 | 0.969 | 0.993 | CNN + mi-SVM | 0.988 | 0.989 | 0.988 | 0.988 | 0.998 | CNN + MI-SVM | 0.982 | 0.982 | 0.982 | 0.982 | 0.996 | -means | 0.991 | 0.991 | 0.991 | 0.991 | 0.998 | Random | 0.976 | 0.976 | 0.976 | 0.975 | 0.995 |
|
|