Data Anomaly Detection through Semisupervised Learning Aided by Customised Data Augmentation Techniques

Table 1

DA techniques for time series in this study.

Data transformation	Description
Jittering	Adding random noise
Scaling	Multiplying a random scalar
Magnitude warping	Convolving a smooth curve varying around one
Time warping	Perturbing temporal locations smoothly
Permutation	Permuting data segments randomly
Random sampling	Subsampling and interpolating