Faster K-Means Cluster Estimation

K-means is a widely used iterative clustering algorithm. There has been considerable work on improving k-means in terms of mean squared error (MSE) and speed, both. However, most of the k-means variants tend to compute distance of each data point to each cluster centroid for every iteration. We propose two heuristics to overcome this bottleneck and speed up k-means. Our first heuristic predicts the candidate clusters for each data point by looking at nearby clusters after first iteration of k-means. Our second heuristic further reduces this candidate cluster list aggressively. We augment well known variants of k-means with our heuristics to demonstrate effectiveness of our heuristics. For various synthetic and real-world datasets, our heuristics achieve speed-up of up-to 10 times without significant increase in MSE.