Showing 1–6 of 6 results for author: Kutuzov, N

Median Clip** for Zeroth-order Non-Smooth Convex Optimization and Multi Arm Bandit Problem with Heavy-tailed Symmetric Noise

Authors: Nikita Kornilov, Yuriy Dorn, Aleksandr Lobanov, Nikolay Kutuzov, Innokentiy Shibaev, Eduard Gorbunov, Alexander Gasnikov, Alexander Nazin

Abstract: In this paper, we consider non-smooth convex optimization with a zeroth-order oracle corrupted by symmetric stochastic noise. Unlike the existing high-probability results requiring the noise to have bounded $κ$-th moment with $κ\in (1,2]$, our results allow even heavier noise with any $κ> 0$, e.g., the noise distribution can have unbounded $1$-st moment. Moreover, our results match the best-known… ▽ More In this paper, we consider non-smooth convex optimization with a zeroth-order oracle corrupted by symmetric stochastic noise. Unlike the existing high-probability results requiring the noise to have bounded $κ$-th moment with $κ\in (1,2]$, our results allow even heavier noise with any $κ> 0$, e.g., the noise distribution can have unbounded $1$-st moment. Moreover, our results match the best-known ones for the case of the bounded variance. To achieve this, we use the mini-batched median estimate of the sampled gradient differences, apply gradient clip** to the result, and plug in the final estimate into the accelerated method. We apply this technique to the stochastic multi-armed bandit problem with heavy-tailed distribution of rewards and achieve $O(\sqrt{Td})$ regret by incorporating noise symmetry. △ Less

Submitted 25 May, 2024; v1 submitted 4 February, 2024; originally announced February 2024.

Breaking the Heavy-Tailed Noise Barrier in Stochastic Optimization Problems

Authors: Nikita Puchkin, Eduard Gorbunov, Nikolay Kutuzov, Alexander Gasnikov

Abstract: We consider stochastic optimization problems with heavy-tailed noise with structured density. For such problems, we show that it is possible to get faster rates of convergence than $\mathcal{O}(K^{-2(α- 1)/α})$, when the stochastic gradients have finite moments of order $α\in (1, 2]$. In particular, our analysis allows the noise norm to have an unbounded expectation. To achieve these results, we s… ▽ More We consider stochastic optimization problems with heavy-tailed noise with structured density. For such problems, we show that it is possible to get faster rates of convergence than $\mathcal{O}(K^{-2(α- 1)/α})$, when the stochastic gradients have finite moments of order $α\in (1, 2]$. In particular, our analysis allows the noise norm to have an unbounded expectation. To achieve these results, we stabilize stochastic gradients, using smoothed medians of means. We prove that the resulting estimates have negligible bias and controllable variance. This allows us to carefully incorporate them into clipped-SGD and clipped-SSTM and derive new high-probability complexity bounds in the considered setup. △ Less

Submitted 17 April, 2024; v1 submitted 7 November, 2023; originally announced November 2023.

Comments: AISTATS 2024. 60 pages, 3 figures. Changes in V2: small typos were fixed, extra experiments and discussion were added. Code: https://github.com/Kutuz4/AISTATS2024_SMoM

Implicitly normalized forecaster with clip** for linear and non-linear heavy-tailed multi-armed bandits

Authors: Yuriy Dorn, Nikita Kornilov, Nikolay Kutuzov, Alexander Nazin, Eduard Gorbunov, Alexander Gasnikov

Abstract: The Implicitly Normalized Forecaster (INF) algorithm is considered to be an optimal solution for adversarial multi-armed bandit (MAB) problems. However, most of the existing complexity results for INF rely on restrictive assumptions, such as bounded rewards. Recently, a related algorithm was proposed that works for both adversarial and stochastic heavy-tailed MAB settings. However, this algorithm… ▽ More The Implicitly Normalized Forecaster (INF) algorithm is considered to be an optimal solution for adversarial multi-armed bandit (MAB) problems. However, most of the existing complexity results for INF rely on restrictive assumptions, such as bounded rewards. Recently, a related algorithm was proposed that works for both adversarial and stochastic heavy-tailed MAB settings. However, this algorithm fails to fully exploit the available data. In this paper, we propose a new version of INF called the Implicitly Normalized Forecaster with clip** (INF-clip) for MAB problems with heavy-tailed reward distributions. We establish convergence results under mild assumptions on the rewards distribution and demonstrate that INF-clip is optimal for linear heavy-tailed stochastic MAB problems and works well for non-linear ones. Furthermore, we show that INF-clip outperforms the best-of-both-worlds algorithm in cases where it is difficult to distinguish between different arms. △ Less

Submitted 26 December, 2023; v1 submitted 11 May, 2023; originally announced May 2023.

Deep Learning-Assisted Localisation of Nanoparticles in synthetically generated two-photon microscopy images

Authors: Rasmus Netterstrøm, Nikolay Kutuzov, Sune Darkner, Maurits Jørring Pallesen, Martin Johannes Lauritzen, Kenny Erleben, Francois Lauze

Abstract: Tracking single molecules is instrumental for quantifying the transport of molecules and nanoparticles in biological samples, e.g., in brain drug delivery studies. Existing intensity-based localisation methods are not developed for imaging with a scanning microscope, typically used for in vivo imaging. Low signal-to-noise ratios, movement of molecules out-of-focus, and high motion blur on images r… ▽ More Tracking single molecules is instrumental for quantifying the transport of molecules and nanoparticles in biological samples, e.g., in brain drug delivery studies. Existing intensity-based localisation methods are not developed for imaging with a scanning microscope, typically used for in vivo imaging. Low signal-to-noise ratios, movement of molecules out-of-focus, and high motion blur on images recorded with scanning two-photon microscopy (2PM) in vivo pose a challenge to the accurate localisation of molecules. Using data-driven models is challenging due to low data volumes, typical for in vivo experiments. We developed a 2PM image simulator to supplement scarce training data. The simulator mimics realistic motion blur, background fluorescence, and shot noise observed in vivo imaging. Training a data-driven model with simulated data improves localisation quality in simulated images and shows why intensity-based methods fail. △ Less

Submitted 17 March, 2023; originally announced March 2023.

$\mathcal{P}$, $\mathcal{T}$-odd Faraday rotation in intracavity absorption spectroscopy with molecular beam as a possible way to improve the sensitivity of the search for the time reflection noninvariant effects in nature

Authors: D. V. Chubukov, L. V. Skripnikov, A. N. Petrov, V. N. Kutuzov, L. N. Labzowsky

Abstract: The present constraint on the space parity ($\mathcal{P}$) and time reflection invariance ($\mathcal{T}$) violating electron electric dipole moment ($e$EDM) is based on the observation of the electron spin precession in an external electric field using the ThO molecule. We propose an alternative approach: observation of the $\mathcal{P}$,~$\mathcal{T}$-odd Faraday effect in an external electric fi… ▽ More The present constraint on the space parity ($\mathcal{P}$) and time reflection invariance ($\mathcal{T}$) violating electron electric dipole moment ($e$EDM) is based on the observation of the electron spin precession in an external electric field using the ThO molecule. We propose an alternative approach: observation of the $\mathcal{P}$,~$\mathcal{T}$-odd Faraday effect in an external electric field using the cavity-enhanced polarimetric scheme in combination with a molecular beam crossing the cavity. Our theoretical simulation of the proposed experiment with the PbF and ThO molecular beams shows that the present constraint on the $e$EDM in principle can be improved by a few orders of magnitude. △ Less

Submitted 9 February, 2021; originally announced February 2021.

Comments: arXiv admin note: text overlap with arXiv:1907.11761

Journal ref: Phys. Rev. A 103, 042802 (2021)

Accurate Evaluation of $\mathcal{P}$,$\mathcal{T}$-odd Faraday Effect in Atoms of Xe and Hg

Authors: D. V. Chubukov, L. V. Skripnikov, L. N. Labzowsky, V. N. Kutuzov, S. D. Chekhovskoi

Abstract: Accurate evaluation of the $\mathcal{P}$,$\mathcal{T}$-odd Faraday effect (rotation of the polarization plane for the light propagating through a medium in presence of an external electric field) is presented. This effect can arise only due to the $\mathcal{P}$,$\mathcal{T}$-odd interactions and is different from the ordinary Faraday effect, i.e. the light polarization plane rotation in an externa… ▽ More Accurate evaluation of the $\mathcal{P}$,$\mathcal{T}$-odd Faraday effect (rotation of the polarization plane for the light propagating through a medium in presence of an external electric field) is presented. This effect can arise only due to the $\mathcal{P}$,$\mathcal{T}$-odd interactions and is different from the ordinary Faraday effect, i.e. the light polarization plane rotation in an external magnetic field. The rotation angle is evaluated for the ICAS (intracavity absorption spectroscopy) type experiments with Xe and Hg atoms. The results show that Hg atom may become a good candidate for a search for the $\mathcal{P}$,$\mathcal{T}$-odd effects in atomic physics. △ Less

Submitted 28 April, 2019; v1 submitted 10 December, 2018; originally announced December 2018.

Journal ref: Phys. Rev. A 99, 052515 (2019)