Research Article
Gesture Recognition from Data Streams of Human Motion Sensor Using Accelerated PSO Swarm Search Feature Selection Algorithm
Table 5
Comparison of gain in accuracy per preprocessing second for different algorithms.
| | FS-PSO | FS-APSO |
| Traditional Classifier | | | HyperPipe (HP) | 0.1721 | 0.3442 | Naive Bayes (NB) | 1.4789333 | 1.3896889 | BayesNet (BN) | 0.1687412 | 0.3298917 | Decision tree (DT) | 0.0446222 | 0.09466 | Random forest (RF) | 0.0084353 | 0.0758806 | Support vector machine (SVM) | 0.0024659 | 0.006978 | Neural network (NN) | −0.0001431 | | Average: | 0.2678793 | 0.3201961 |
| Incremental classifier | | | Random Hoeffding tree (RHT) | 2.32932 | 2.6506 | Hoeffding tree (HT) | −0.484075 | 0.0688467 | Hoeffding pption tree (HOT) | 1.0483455 | 1.26219 | NBUpdateable (NBup) | 1.1729444 | 1.027925 | Active (AC) | 0.0052164 | 0.049342 | Adwin (AW) | −0.016625 | 0.0109092 | Kstar | 0.0055993 | 0.0075781 | Average: | 0.5801037 | 0.7253416 |
|
|