Draft:Aleksander Mądry
![]() | Draft article not currently submitted for review.
This is a draft Articles for creation (AfC) submission. It is not currently pending review. While there are no deadlines, abandoned drafts may be deleted after six months. To edit the draft click on the "Edit" tab at the top of the window. To be accepted, a draft should:
It is strongly discouraged to write about yourself, your business or employer. If you do so, you must declare it. Where to get help
How to improve a draft
You can also browse Wikipedia:Featured articles and Wikipedia:Good articles to find examples of Wikipedia's best writing on topics similar to your proposed article. Improving your odds of a speedy review To improve your odds of a faster review, tag your draft with relevant WikiProject tags using the button below. This will let reviewers know a new draft has been submitted in their area of interest. For instance, if you wrote about a female astronomer, you would want to add the Biography, Astronomy, and Women scientists tags. Editor resources
Last edited by 2401hz (talk | contribs) 5 months ago. (Update) |
Aleksander Mądry | |
---|---|
Alma mater | Massachusetts Institute of Technology (PhD) |
Scientific career | |
Fields | Computer Science |
Institutions | Massachusetts Institute of Technology |
Thesis | From Graphs to Matrices, and Back: New Techniques for Graph Algorithms (2011) |
Doctoral advisor | Michel Xavier Goemans and Jonathan A. Kelner |
Aleksander Mądry is a computer scientist and professor at the Massachusetts Institute of Technology. As of 2024 he is on leave at OpenAI leading preparedness team.
Biography
Aleksander Mądry was born in Wrocław, Poland and obtained a bachelor's degree in theoretical physics from the University of Wrocław. He continued his studies at the Massachusetts Institute of Technology with an M.S. in computer science ("Faster Generation of Random Spanning Trees") and a Ph.D. in 2011 [2] under the supervision of Michel Xavier Goemans and Jonathan A. Kelner ("From Graphs to Matrices, and Back: New Techniques for Graph Algorithms"). He spent a year as a postdoc at Microsoft Research New England, then he worked at the École Polytechnique Fédérale de Lausanne (EPFL) until 2015, when he joined the department of electrical engineering and computer science at MIT.
In 2023 he took leave at MIT to join OpenAI to head their "Preparedness" team to evaluate the potential of machine learning models to cause "catastrophic" damage.[1][2] On March 8th Mądry testified to congress on the opportunities and risks of AI[3]. After the Novemer 17 removal of Sam Altman from OpenAI, Mądry resigned, before returning after Altman's reinstatement. [4]
Research
Aleksander Mądry's early work made substantial contributions to the theory of algorithms[5]. In 2011 he developed an approximation algorithm for the maximum flow problem, the first improvement in many years. And then, in 2013, he gave an exact calculation algorithm for the maximum flow problem which was the first to improve on the prior bound of Evan and Tarjan in 1975[6]. Mądry also contributed advances in the k-server problem and the traveling salesman problem [7]. The Pressburger Award wrote: "Aleksander’s results have been celebrated in the community not only because he broke long standing complexity barriers but moreover because he introduced new and very different techniques to the field which since have successfully been picked up by others."
Mądry's most recent work focuses on machine learning and artificial intelligence. In 2018 he developed the method of "adversarial training", a scheme to improve the robustness of machine learning models[8]. Refinements of adversarial training remain the best methods for improving adversarial robustness. Since 2018 Mądry has extended his work to cover other areas of explaining why machine learning works, and investigating the role of data in the accuracy of machine learning models.
Awards
- Pressburger Award 2018. (https://eatcs.org/index.php/presburger)
- FOCS 2013. Best Paper Award (https://madry.mit.edu/docs/matching_flow.pdf)
- FOCS 2011. Best Paper Award (https://madry.mit.edu/docs/kserver.pdf)
- STOC 2011. Best Paper Award (https://madry.mit.edu/docs/maxflow.pdf)
- SODA 2011. Best Paper Award (https://madry.mit.edu/docs/atsp.pdf)
References
- ^ Roth, Emma (2023-10-26). "OpenAI forms new team to assess "catastrophic risks" of AI". The Verge. Retrieved 2024-03-04.
- ^ "Frontier risk and preparedness". openai.com. Retrieved 2024-03-04.
- ^ Madry, Aleksander (2023). "Written Statement of Aleksander Mądry" (PDF).
- ^ "OpenAI Staff Threatens Exodus, Jeopardizing Company's Future". 2023-11-20. Retrieved 2024-03-04.
- ^ "The Presburger Award 2018 Laudatio for Aleksander Madry" (PDF).
- ^ "Navigating Central Path with Electrical Flows: From Flows to Matchings, and Back | IEEE Conference Publication | IEEE Xplore". ieeexplore.ieee.org. Retrieved 2024-03-04.
- ^ Asadpour, Arash; Goemans, Michel X.; Mądry, Aleksander; Gharan, Shayan Oveis; Saberi, Amin (2010-01-17). "An O (log n /log log n )-approximation Algorithm for the Asymmetric Traveling Salesman Problem". Society for Industrial and Applied Mathematics: 379–389. doi:10.1137/1.9781611973075.32. ISBN 978-0-89871-701-3.
{{cite journal}}
: Cite journal requires|journal=
(help) - ^ Madry, Aleksander; Makelov, Aleksandar; Schmidt, Ludwig; Tsipras, Dimitris; Vladu, Adrian (2019-09-04), Towards Deep Learning Models Resistant to Adversarial Attacks, doi:10.48550/arXiv.1706.06083, retrieved 2024-03-04