Piotr Indyk, Ph.D.

Piotr Indyk is noted for his work on efficient approximate algorithms for high-dimensional geometric problems. This includes the nearest neighbor search, where given a data point, the goal is to find points highly similar to it without scanning the whole data set. To address this problem, he co-developed the technique of locality sensitive hashing, which proved to be influential in many applications, ranging from data mining to computer vision. He has also made significant contributions to sublinear algorithms for massive data problems. In particular, he has developed several approximate algorithms for massive data streams that use very limited space. Recently, he has co-developed new algorithms for the sparse Fourier transform, which compute the Fourier transform of signals with sparse spectra faster than the FFT algorithm.