Brief
UBC researchers receive Test of Time Award for gradient‑descent paper
The team’s 2016 work on the Polyak‑Lojasiewicz condition earned the Test of Time Award at the European Conference on Machine Learning in Naples.
By Felo News Desk · Published
University of British Columbia computer‑science professor Mark Schmidt and former trainees Hamed Karimi and Julie Nutini were presented with the Test of Time Award at the 2026 European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD) in Naples, Italy.
The award recognises their 2016 paper that introduced a simple condition – now called the Polyak‑Lojasiewicz condition – enabling a three‑line proof of fast convergence for gradient‑descent methods. The paper has been cited over 2,000 times and has informed guarantees for most standard optimisation algorithms, as well as newer areas such as reinforcement learning and diffusion‑model sampling.
Karimi, now a senior applied scientist at Amazon, said the recognition a decade later “means a lot, especially shared with Julie and Mark and rooted in my time at UBC.” Schmidt added that the work stemmed from his frustration that existing textbook results applied only to strongly convex functions, prompting the search for a more realistic assumption.
Key facts
- Mark Schmidt, Hamed Karimi and Julie Nutini received the Test of Time Award at ECML PKDD 2026. (cs.ubc.ca)
- The award honoured their 2016 paper on the Polyak‑Lojasiewicz condition for fast gradient‑descent convergence. (cs.ubc.ca)
- The paper has been cited over 2,000 times and influenced optimisation research across multiple domains. (cs.ubc.ca)
Sources
- [1] cs.ubc.ca — originally reported as “UBC Computer Science researchers win Test of Time Award for influential paper at leading machine learning conference”
