Incremental Constrained Random Walk Clustering

In many real-world application scenarios, data usually incrementally update over time. In such cases, traditional constrained clustering algorithms become unsuitable for dealing with incremental data because of high computational cost. In this paper, we propose a novel incremental constrained random walk clustering algorithm (ICC), which not only efficiently deal with the incremental data but also utilize the incremental constraints. To reduce the time complexity, it updates the influence range of each selected data point and utilizes the intermediate structure of the previous time step. Extensive experiment results on datasets demonstrate that our algorithm is both effective and efficient.