Algorithm 1 Training phase

Input: raw dataset without annotation $Y=\{y_i|i=1,\dots ,M\}$, prelabeled dataset $X^p=\{x_i^p|i=1,\dots ,N\},C^p=\{c_k^p|k=1,\dots ,K\},S_k^p=\{x_j^{S_k^p}|j=1,\dots ,n^{S_k}\}$, max iterations T, nearest neighbor number k, outlier threshold ${\epsilon }_1$

Output: personalized MGD models $p_1\left(x\right),p_2\left(x\right),\mathrm{and}{p}_3\left(x\right)$ Preprocess the dataset $Y=\left\{y_i\right|i=1,\dots ,M\}$ through the Butterworth low-pass filter and dividing by sensitivity coefficient $k$. Segment the data into overlapping windows. Extract the features and normalization. Save the results from step 1 and step 2 as $X=\left\{x_i\right|i=1,\dots ,M\}$. Initialize K-Means clustering centroids $c_k^0$ by Equation (3). For t = 1 to T do the following:  Assign samples that belong to the kth cluster $S_k$ by Equation (4).  Update kth centroids $c_k^{\left(t+1\right)}$ by Equation (5).  If there is no change in the assignment step   break  end if end for For i = 1 to M do the following:  $scor{e}_i=LLOF\left({\mathrm{x}}_{\mathrm{i}}\right),{\mathrm{x}}_{\mathrm{i}}\in X$  if $scor{e}_i>{\epsilon }_1$   remove $x_i$ from $X$  end if end for Using Equation (9) and (10), compute ${\mu }_1,{\Sigma }_1$ by the LIA cluster $S_1=\{x_J^{S_1}|j=1,\dots ,m^{S_1}\}$, compute ${\mu }_2,{\Sigma }_2$ by the MIA cluster $S_2=\{x_j^{S_2}|j=1,\dots ,m^{S_2}\}$, and compute ${\mu }_3,{\Sigma }_3$ by the VIA cluster $S_3=\{x_j^{S_3}|j=1,\dots ,m^{S_3}\}$. Establish the personalized MGD models: $p_1\left(x\right)=\frac{1}{{\left(2\pi \right)}^{\frac{n}{2}}{\left|{\Sigma }_1\right|}^{\frac{1}{2}}}exp\left(-\frac{1}{2}{\left(x-{\mu }_1\right)}^T{\Sigma }_3{}^{-1}\left(x-{\mu }_1\right)\right)$, $p_2\left(x\right)=\frac{1}{{\left(2\pi \right)}^{\frac{n}{2}}{\left|{\Sigma }_2\right|}^{\frac{1}{2}}}exp\left(-\frac{1}{2}{\left(x-{\mu }_2\right)}^T{\Sigma }_2{}^{-1}\left(x-{\mu }_2\right)\right)$, and $p_3\left(x\right)=\frac{1}{{\left(2\pi \right)}^{\frac{n}{2}}{\left|{\Sigma }_3\right|}^{\frac{1}{2}}}exp\left(-\frac{1}{2}{\left(x-{\mu }_3\right)}^T{\Sigma }_3{}^{-1}\left(x-{\mu }_3\right)\right)$