|
| Work | Sensor | Features | Algorithm | Traits | Results |
|
| [186] | Accelerometer | Time-domain features | ANN, J48 decision tree algorithms [187], and instance-based learning (IBk) [188] | Weight, height, and gender | 71.2% for gender using IBk, 85.7% for height using ANN, and 78.9% for weight using IBk |
| [182] | Accelerometer and touchscreen | Time-domain features, touch pressure, and size | -mean nearest neighbor | User identification | More than 96% for identification |
| [77] | Touchscreen | Delay between pressing two different keys | ANN, nearest neighbor, SVM, gradient descent bp, Euclidean distance, linear discriminant analysis, and another 5 algorithms | Classifying children from adults | More than 92% for SVM and 89% for linear discriminant analysis |
| [80] | Touchscreen | Delay and duration of pressing | SVM | Gender classification | Accuracy of 91% |
| [81] | Touchscreen, accelerometer, and gyroscope | 29 features including: special keys, total keys pressed, number of backspaces used, edit distance, total completion time, average time between keys | Decision tree (number of keys), SVC linear kernel (age), SVC linear kernel (gender), logistic regression, -nearest, and Gaussian NB | Number of fingers used, gender, and age | 80% for the number of fingers, 75% for age, and 60% for gender |
| [189] | Touchscreen gestures, gyroscope, accelerometer | 14 gesture features, total length, total time, width, height, area, pressure, speed, acceleration, arc distance, and angle start to end | SVM, logistic regression, naive Bayes, J48 | Gender classification | 71% accuracy for logistic regression |
| [190] | Fingerprint | Wavelet features and singular value decomposition | -nearest | Gender classification | Accuracy exceeded 88% |
| [87] | Touchscreen | Swipe gesture speed in four directions and other features from [189] | Statistical | Thumb length and users’ height | Accuracy of 72% of the relation between thumb length and height |
|
