معرفی
John Wright is an Associate Professor in the Department of Electrical Engineering at Columbia University's Fu Foundation School of Engineering and Applied Science, with additional affiliations in the Department of Applied Physics and Applied Mathematics and the Data Science Institute. He maintains his office at 408 Mudd Hall within the Data Science Institute and leads a research group focused on high-dimensional data analysis.
Wright received his PhD in Electrical Engineering from the University of Illinois at Urbana-Champaign in October 2009, followed by a research position at Microsoft Research from 2009-2011 before joining Columbia University. His research focuses on developing algorithms for robustly recovering structured signal representations from incomplete and corrupted observations, with applications in imaging and vision.
His research interests span high-dimensional data analysis, signal processing, computer vision, and optimization. Wright's work places particular emphasis on finding methods that perform well even when data is unreliable - noisy or corrupted - and which come with proofs of correctness. His research group develops tools for robustly analyzing high-dimensional data and applying them to practical problems such as image and video compression, face and object recognition.
Wright's publication record shows a clear evolution from sparse representation and dictionary learning toward more complex deep learning architectures. His recent work focuses on geometric approaches to nonconvex optimization problems, blind deconvolution methods, and the theoretical foundations of deep networks. His publications consistently bridge theoretical guarantees with practical applications across computer vision and signal processing domains.
Scientific Awards and Honors:
- 2009 Lemelson-Illinois Prize for Innovation for his work on face recognition
- 2009 UIUC Martin Award for Excellence in Graduate Research
- 2008-2010 Microsoft Research Fellowship
- 2012 COLT Best Paper Award (with Wang and Spielman)
Wright has successfully mentored numerous PhD students who have gone on to prestigious positions at institutions including TTIC, Google, Rutgers University, and Boston University. He also co-authored the influential textbook 'High-Dimensional Data Analysis with Low-Dimensional Models: Principles, Computation, and Applications' published by Cambridge University Press in 2022, which has been used as course material at multiple leading universities.
His research group maintains active collaborations across multiple institutions and regularly presents at top conferences including NeurIPS, ICML, CVPR, and COLT. The group holds weekly seminars (typically on Thursdays from 10-12 in 408 Mudd) where current research is discussed and developed.
