Efficient Generation of Hidden Outliers for Improved Outlier Detection
This addresses the problem of improving outlier detection performance for data scientists, though it appears incremental as it builds on existing outlier generation techniques.
The paper tackles the challenge of generating realistic outliers for outlier detection by proposing BISECT, a method that efficiently mimics the 'multiple views' property in high-dimensional spaces, reducing error by up to 3 times compared to baselines.
Outlier generation is a popular technique used for solving important outlier detection tasks. Generating outliers with realistic behavior is challenging. Popular existing methods tend to disregard the 'multiple views' property of outliers in high-dimensional spaces. The only existing method accounting for this property falls short in efficiency and effectiveness. We propose BISECT, a new outlier generation method that creates realistic outliers mimicking said property. To do so, BISECT employs a novel proposition introduced in this article stating how to efficiently generate said realistic outliers. Our method has better guarantees and complexity than the current methodology for recreating 'multiple views'. We use the synthetic outliers generated by BISECT to effectively enhance outlier detection in diverse datasets, for multiple use cases. For instance, oversampling with BISECT reduced the error by up to 3 times when compared with the baselines.