Life-long phishing attack detection using continual learning

This section covers a detailed discussion of our methodology to identify and mitigate the performance drop. We analyzed our dataset features using low-dimensional principal component analysis (PCA) embedding of samples to visualize the difference in distribution. We transformed features into 1-D and 2-D PCA as shown in Figs. 1 and 2, respectively. Figure 1 shows the features distribution of 1-D PCA trends for data samples taken from 3 consecutive years, and the x-axis represents 1-D feature values while the y-axis represents the frequency of samples. It depicts the change in the…