Scientists Propose Method for More Efficient Resource Use in Machine Learning
An international group of researchers, including mathematicians from the AI and Digital Science Institute at the HSE Faculty of Computer Science, has provided a theoretical justification for a simple and computationally efficient method of estimating uncertainty in Stochastic Gradient Descent (SGD). The paper has been published on the scientific preprint server arXiv.org and presented at AISTATS 2026.
Stochastic algorithms, including SGD, are widely used in optimisation and machine learning tasks. Because these algorithms incorporate randomness, such as randomly selected mini-batches of data, an important feature of their solutions is the confidence interval—the range within which the true solution is likely to lie. Traditional approaches to constructing such intervals rely on complex statistical estimations, particularly explicit estimates of the solution’s marginal covariance matrix. These methods can be computationally expensive and may still produce inaccurate uncertainty estimates.
A covariance matrix is a table that shows how several random variables, such as features or parameters, are related to one another and how they vary around their mean values.
An international team, including researchers from the AI and Digital Science Institute at the HSE Faculty of Computer Science, analysed an empirically popular approach to estimating confidence intervals for averaged SGD that does not require repeated model training or complex calculations. The authors demonstrated that this method accurately reproduces the distribution of the averaged SGD solution and does not require an explicit estimate of the marginal covariance matrix.
Marina Sheshukova
'Similar methods have already been used in practice and have often demonstrated better results than alternative approaches. We wanted to understand the reasons behind this empirical advantage and were able to provide a rigorous mathematical interpretation,' explained Marina Sheshukova, Junior Research Fellow of the International Laboratory of Stochastic Algorithms and High-Dimensional Inference at the HSE AI and Digital Science Institute.
This mathematical justification makes it possible to reassess simple empirical methods for estimating uncertainty in machine learning. Developers will be able to obtain reliable uncertainty estimates faster and with fewer computational resources. This is particularly important in fields where it is essential to know not only the prediction itself but also the level of confidence associated with it, such as medicine, finance, and autonomous systems.
See also:
HSE Computer Science Researchers Win Gold Medal at International Machine Learning Competition
A team comprising HSE International Laboratory of Statistical and Computational Genomics researchers Aleksei Shmelev and Nikita Chervov, 2025 graduate of the HSE Faculty of Computer Science’s Master’s programme in Data Analysis in Biology and Medicine Ivan Gevorkov, and two students from the United States achieved an outstanding result at the 2026 NeuroGolf international machine learning championship. The team won a gold medal and placed seventh overall.
‘I Would Like My Research to Help Make the World a Calmer and Better Place’
Whatever task Saraa Ali, Junior Research Fellow at the Laboratory of Methods for Big Data Analysis (LAMBDA) of the AI and Digital Science Institute (HSE Faculty of Computer Science), is working on, she thinks about how it can benefit people. She told the Young Scientists of HSE University project about her large family, diagnosing three-phase motors, and her dream of building a children’s home in her native country.
‘The Peak of Stupidity’ and ‘The Valley of Despair’: HSE Economists Propose an Explanation for the Dunning–Kruger Effect
The Dunning–Kruger effect, which describes a sharp surge in self-confidence among beginners followed by an equally rapid decline as they gain experience, can be explained by the nature of the learning process and the acquisition of new knowledge. This conclusion was reached by Andrey Vorchik of the HSE Faculty of Economic Sciences together with independent researcher Murat Mamyshev. They developed a mathematical model of learning and demonstrated how subjective confidence is formed and changes as knowledge accumulates, as well as how teachers can reduce the ‘valley of despair’ experienced by learners.
Toffee and Risk: Scientists Discover Why People Who Crave Sweets Make More Impulsive Choices
Having a sweet tooth may be linked not only to eating habits but also to the way people make decisions. Researchers at HSE University have found that people with a preference for sweet foods tend to behave more impulsively—not because they want immediate rewards, but because they are less willing to tolerate uncertainty. These findings may help improve treatments for addiction. The study findings have been published in Frontiers in Psychology.
Physicists Find a Way to Model Ion Parameters in Plasma in Seconds
Researchers from HSE University and the Moscow Institute of Physics and Technology (MIPT) have developed a set of simple analytical methods for calculating the properties of heavy ions in helium under the influence of a strong electric field. The new approach speeds up calculations of ion mobility and ion–molecule reaction rates by thousands of times while maintaining sufficient accuracy for plasma jet modelling. The findings have been published in the journal Physica Scripta.
Two Years of Growth or Decline: How to Choose an Investment Strategy
Economists from HSE University, together with colleagues from international universities, have analysed stock market movements over almost a century and proposed an investment strategy that could have delivered returns nearly twice as high as the market average. Their research suggests following a momentum strategy during periods of sustained market growth and switching to a value strategy after prolonged market declines. The study has been published in the Journal of Banking and Finance.
Researchers Reveal Link Between Attention and Communication Difficulties in Autism
Researchers at HSE University have examined how communication difficulties in children with autism are related to brain function. The findings show that not only language networks but also attention networks play an important role. The weaker the connections involved in maintaining focus and switching attention, the more pronounced communication difficulties were. The study has been published in European Child & Adolescent Psychiatry.
Scientists Discover Why Some People Wore Masks During COVID-19 While Others Did Not
Why do some people voluntarily follow new rules while others ignore them? Researchers at HSE University have found that the answer lies not so much in people's willingness to cooperate, as previously believed, but in their ability to empathise with others. Empathy proved to be the strongest predictor of whether people chose to wear face masks voluntarily during the COVID-19 pandemic. The findings have been published in Frontiers.
Speed, Precision, and Self-Correction: HSE Faculty of Computer Science Researchers at ICML-2026
Researchers from the HSE Faculty of Computer Science (FCS) presented their work at theInternational Conference on Machine Learning (ICML 2026) in Seoul, South Korea, one of the leading scientific events in the field. Several projects by the faculty’s researchers received the prestigious Spotlight distinction.
Physicists at HSE University and FIAN Discover Way to 'Photograph' Sound for Testing Materials Used in 6G Communications
Researchers at HSE University, in collaboration with colleagues from the Lebedev Physical Institute of the Russian Academy of Sciences (FIAN), have developed a method for rapidly determining how firmly a film is bonded to a substrate. This is important for the creation of ultrahigh-frequency acoustic filters, which are key components of next-generation 5G and 6G communications. For the first time, researchers have succeeded in measuring the lateral rigidity of the bond between a two-dimensional material film and a substrate in this way. The study results have been published in Applied Physics Letters.