Fig. 3
From: A novel algorithm for fast and scalable subspace clustering of high-dimensional data
Numerical experiments for probability of collisions. A trial includes drawing a label set of 4 random integers with the same number of digits e.g., {333,444,555,666} is a sample set where no. of digits is 3. The probability of drawing same set of integers reduces drastically with the use of larger integers as labels in the set, e.g., ≈ 10 collisions for 100 million trials when the label is a 12-digit integer