A5036951521- Sparse and Compressive Sensing Techniques
- Stochastic Gradient Optimization Techniques
- COVID-19 diagnosis using AI
- Fluid Dynamics and Turbulent Flows
- Advanced Optimization Algorithms Research
- Markov Chains and Monte Carlo Methods
- Artificial Intelligence in Healthcare and Education
- Privacy-Preserving Technologies in Data
- Blind Source Separation Techniques
- Radiomics and Machine Learning in Medical Imaging
- Tensor decomposition and applications
- Statistical and numerical algorithms
- Numerical Methods and Algorithms
- Numerical methods in inverse problems
- Geomagnetism and Paleomagnetism Studies
- Plant Water Relations and Carbon Dynamics
- Multilingual Education and Policy
- Gender Studies in Language
- Anomaly Detection Techniques and Applications
- Second Language Learning and Teaching
- Microwave Imaging and Scattering Analysis
- Nanofluid Flow and Heat Transfer
- Computational Physics and Python Applications
- Language, Discourse, Communication Strategies
- Discourse Analysis in Language Studies
University of Cambridge
2019-2024
University of Oxford
2022
Lung Institute
2021
AstraZeneca (United Kingdom)
2021
Cambridge Hospital
2021
Applied Mathematics (United States)
2019
University of Colorado Boulder
2015
Machine learning methods offer great promise for fast and accurate detection prognostication of COVID-19 from standard-of-care chest radiographs (CXR) computed tomography (CT) images. Many articles have been published in 2020 describing new machine learning-based models both these tasks, but it is unclear which are potential clinical utility. In this systematic review, we search EMBASE via OVID, MEDLINE PubMed, bioRxiv, medRxiv arXiv papers preprints uploaded January 1, to October 3,...
Related DatabasesWeb of Science You must be logged in with an active subscription to view this.Article DataHistorySubmitted: 21 December 2020Accepted: 20 July 2021Published online: 2021Keywordsnonconvex and nonsmooth optimization, stochastic variance reduction, Kurdyka--Łojasiewicz inequality, PALMAMS Subject Headings90C26, 90C15, 90C30, 49M27Publication DataISSN (online): 1936-4954Publisher: Society for Industrial Applied MathematicsCODEN: sjisbi
The influence of fixed temperature and heat flux thermal boundary conditions on rapidly rotating convection in the plane layer geometry is investigated for case stress-free mechanical conditions. It shown that whereas leading-order system satisfies implicitly, a double structure necessary to satisfy layers consist classical Ekman adjacent solid boundaries adjust viscous stresses zero, wind balance just outside adjusts normal derivative fluctuation zero. these interior geostrophically...
Abstract Variance reduction is a crucial tool for improving the slow convergence of stochastic gradient descent. Only few variance-reduced methods, however, have yet been shown to directly benefit from Nesterov’s acceleration techniques match rates accelerated methods. Such approaches rely on “negative momentum”, technique further variance that generally specific SVRG estimator. In this work, we show first time negative momentum unnecessary and develop universal framework allows all popular...
We introduce SPRING, a novel stochastic proximal alternating linearized minimization algorithm for solving class of non-smooth and non-convex optimization problems. Large-scale imaging problems are becoming increasingly prevalent due to advances in data acquisition computational capabilities. Motivated by the success methods, we propose variant (PALM) \cite{bolte2014proximal}. provide global convergence guarantees, demonstrating that our proposed method with variance-reduced gradient...
This paper explores how textbooks for second language learners make ideologically motivated decisions regarding which forms of the target are included in curriculum. We explore this question via an analysis Korean textbooks, at a time when society is becoming increasingly globalized and multicultural. The focus on representation highly developed system address terms Korean. Although known to play important role shaping intimacy hierarchy interactions, represent them vastly simplified ways....
We present a uniform analysis of biased stochastic gradient methods for minimizing convex, strongly and non-convex composite objectives, identify settings where bias is useful in estimation. The framework we allows us to extend proximal support algorithms, including SAG SARAH, the first time convex setting. also use our develop new algorithm, Stochastic Average Recursive GradiEnt (SARGE), that achieves oracle complexity lower-bound non-convex, finite-sum objectives requires strictly fewer...
We introduce a reformulation of regularized low-rank recovery models to take advantage GPU, multiple CPU, and hybridized architectures. Low-rank often involves nuclear norm minimization through iterative thresholding singular values. These are slow fit difficult parallelize because their dependence on computing value decomposition at each iteration. Regularized also incorporate nonsmooth terms separate structured components (e.g., sparse outliers) from the component, making these problems...
This paper studies tensor-based Robust Principal Component Analysis (RPCA) using atomic-norm regularization. Given the superposition of a sparse and low-rank tensor, we present conditions under which it is possible to exactly recover components. Our results improve on existing performance guarantees for tensor-RPCA, including those matrix RPCA. also show that regularization provides better recovery tensor-structured data sets than other approaches based matricization. In addition these...
The Condat-V\~u algorithm is a widely used primal-dual method for optimizing composite objectives of three functions. Several algorithms two functions are special cases Condat-V\~u, including proximal gradient descent (PGD). It well-known that PGD exhibits suboptimal performance, and simple adjustment to can accelerate its convergence rate from $\mathcal{O}(1/T)$ $\mathcal{O}(1/T^2)$ on convex objectives, this accelerated optimal. In work, we show the allows it recover (APGD) as case,...
We consider the task of image reconstruction while simultaneously decomposing reconstructed into components with different features. A commonly used tool for this is a variational approach an infimal convolution appropriate functions as regularizer. Especially noise corrupted observations, incorporating these functionals classical method Bregman iterations provides robust obtaining overall good approximation true image, by stopping early iteration according to discrepancy principle. However,...
We introduce AnySOS, an implementation of algorithm Renegar for solving sum-of-squares (SOS) programs arising in control applications. Renegar's is efficient first-order method general semidefinite programs. One its key features, unlike other methods, that it produces feasible iterates throughout execution. This particularly important applications: the can be stopped anytime and still return a valid certificate/proof underlying system. For critical real-time applications this used to...
Buoyancy-driven turbulence of liquid metal is the primary driving-force planetary magnetic fields.Compositional convection characterized by density heterogeneities inducing a buoyant force in metallic fluids that compose planet's interior.For rotating compositional convection, dependent parameters are Schmidt number (Sc = ν/D), Rayleigh (Ra gρH 4 /Dν), and Ekman (E ν/2ΩH 2 ).Asymptotically reducing governing equations limit E 0, we able to perform direct numerical simulations highly...
The influence of fixed temperature and heat flux thermal boundary conditions on rapidly rotating convection in the plane layer geometry is investigated for case stress-free mechanical conditions. It shown that whereas leading order system satisfies implicitly, a double structure necessary to satisfy layers consist classical Ekman adjacent solid boundaries adjust viscous stresses zero, wind balance just outside adjusts such are satisfied. these interior geostrophically balanced be...