David Donoho

David Donoho developed key wavelet-based methods for signal denoising and compression, including wavelet thresholding techniques that became standard tools in signal processing. His work, often in collaboration with Iain Johnstone, showed that wavelet thresholding achieves near-optimal recovery of signals from noisy data.

Donoho was one of the key researchers who formalized compressed sensing theory in the mid-2000s, alongside Emmanuel Candès and Terence Tao. He proved that sparse signals could be recovered from far fewer linear measurements than the Nyquist-Shannon sampling theorem would require, using l1-minimization.

David Donoho is a professor of Statistics and Electrical Engineering at Stanford University. He earned his PhD from Harvard under David Freedman and has been at Stanford since the 1990s.

Donoho wrote the widely cited essay '50 Years of Data Science' (2015), in which he argued for the importance of reproducible research and criticized some modern data science practices. He has long advocated for literate computing tools like Jupyter notebooks to ensure scientific transparency.