Table 14 Comparative analysis of CoLog performance across various window sizes and sequence types within the Jhuisi dataset. Bold numbers indicate the outstanding results.

From: A unified framework for detecting point and collective anomalies in operating system logs via collaborative transformers

Window size

Sequence type

Background

Context

1

2

3

6

9

1

2

3

6

9

Precision

99.97

99.76

99.94

99.88

99.78

\(\textbf{100}\)

99.92

99.52

99.85

99.75

Recall

99.91

99.76

99.81

99.88

99.25

\(\textbf{100}\)

99.72

99.52

99.79

99.16

F1-Score

99.94

99.76

99.88

99.88

99.51

\(\textbf{100}\)

99.82

99.52

99.82

99.45

Accuracy

99.96

99.83

99.91

99.91

99.66

\(\textbf{100}\)

99.87

99.66

99.87

99.62