Optimal Quantile Estimation for Streams
Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only! Grab it Explore a 38-minute lecture on optimal quantile estimation for data streams presented by Mihir Singhal from UC Berkeley at the Simons Institute. Delve into the fundamental problem of data sketching, focusing on estimating quantiles in a stream of elements. Learn about previous algorithms, including comparison-based methods and their space complexity. Discover a new deterministic quantile sketch that achieves optimal memory usage of O(1/ε) words, improving upon existing algorithms. Understand the significance of this advancement in approximating stream statistics like median or percentiles with minimal space requirements. Gain insights into the collaborative work with Meghal Gupta and Hongxun Wu, which contributes to the field of extroverted sublinear algorithms.