检索结果-内蒙古大学图书馆

Improved Sparse adaptive algorithms for Accurate Non-contact Heartbeat Detection Using Time-Window-Variation Technique

学校读者我要写书评

暂无评论

Improved Sparse Adaptive Algorithms for Accurate Non-contact...

Annual International Conference of the IEEE Engineering in Medicine and Biology Society

作者： Chen Ye Kentaroh Toyoda Tomoaki Ohtsuki Keio Gijuku Daigaku Minato-ku Tokyo JP

Recently, a sparse adaptive algorithm termed zero-attracting sign least-mean-square (ZA-SLMS), has been clarified to be able to reconstruct robustly heartbeat spectrum by Doppler radar signal. However, since the strengths of noise evidently differ under different body motions, the sparse heartbeat spectra cannot be always acquired accurately by the constant regularization parameter (REPA) that balances the gradient correction and the sparse penalty, applying in the ZA-SLMS algorithm. In this paper, an improved ZA-SLMS algorithm is proposed by introducing adaptive REPA (AREPA), where the proportion of sparse penalty is adjusted based on the standard deviation of radar data. Moreover, to enhance the stability of heartbeat detection, a time-window-variation (TWV) technique is further introduced in the improved ZA-SLMS algorithm, considering the fact that the position of spectral peak associated with the heart rate (HR) is stable when the length of time window changes within a short period. Experimental results measured against five subjects validated that our proposal reliably improves the error of HR estimation than the standard ZA-SLMS algorithm.

关键词： Heart rate variability Doppler radar Heart beat Estimation adaptive algorithms

Polygonal Unadjusted Langevin algorithms: Creating stable and efficient adaptive algorithms for neural networks

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Lim, Dong-Young Sabanis, Sotirios Department of Industrial Engineering UNIST Korea Republic of School of Mathematics The University of Edinburgh United Kingdom The Alan Turing Institute United Kingdom National Technical University of Athens Greece

We present a new class of Langevin-based algorithms, which overcomes many of the known shortcomings of popular adaptive optimizers that are currently used for the fine tuning of deep learning models. Its underpinning theory relies on recent advances of Euler-Krylov polygonal approximations for stochastic differential equations (SDEs) with monotone coefficients. As a result, it inherits the stability properties of tamed algorithms, while it addresses other known issues, e.g. vanishing gradients in deep learning. In particular, we provide a nonasymptotic analysis and full theoretical guarantees for the convergence properties of an algorithm of this novel class, which we named THΕO POULA (or, simply, TheoPouLa). Finally, several experiments are presented with different types of deep learning models, which show the superior performance of TheoPouLa over many popular adaptive optimization algorithms. Copyright © 2021, The Authors. All rights reserved.

关键词： adaptive algorithms

MaxGap bandit: adaptive algorithms for approximate ranking

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Katariya, Sumeet Tripathy, Ardhendu Nowak, Robert University of Wisconsin MadisonWI53706 United States

This paper studies the problem of adaptively sampling from K distributions (arms) in order to identify the largest gap between any two adjacent means. We call this the MaxGap-bandit problem. This problem arises naturally in approximate ranking, noisy sorting, outlier detection, and top-arm identification in bandits. The key novelty of the MaxGap bandit problem is that it aims to adaptively determine the natural partitioning of the distributions into a subset with larger means and a subset with smaller means, where the split is determined by the largest gap rather than a pre-specified rank or threshold. Estimating an arm's gap requires sampling its neighboring arms in addition to itself, and this dependence results in a novel hardness parameter that characterizes the sample complexity of the problem. We propose elimination and UCB-style algorithms and show that they are minimax optimal. Our experiments show that the UCB-style algorithms require 6-8x fewer samples than non-adaptive sampling to achieve the same error. Copyright © 2019, The Authors. All rights reserved.

关键词： adaptive algorithms

Uncertainty about uncertainty: Optimal adaptive algorithms for estimating mixtures of unknown coins∗

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Lee, Jasper C.H. Valiant, Paul Brown University United States IAS and Purdue University

Given a mixture between two populations of coins, "positive" coins that each have—unknown and potentially different—bias ≥ 12 + ∆ and "negative" coins with bias ≤ 21 − ∆, we consider the task of estimating the fraction ρ of positive coins to within additive error Ε. We achieve an upper and lower bound of Θ(Ε2ρ∆2 log 1δ) samples for a 1 − δ probability of success, where crucially, our lower bound applies to all fully-adaptive algorithms. Thus, our sample complexity bounds have tight dependence for every relevant problem parameter. A crucial component of our lower bound proof is a decomposition lemma (see Lemmas 17 and 18) showing how to assemble partially-adaptive bounds into a fully-adaptive bound, which may be of independent interest: though we invoke it for the special case of Bernoulli random variables (coins), it applies to general distributions. We present simulation results to demonstrate the practical efficacy of our approach for realistic problem parameters for crowdsourcing applications, focusing on the "rare events" regime where ρ is small. The fine-grained adaptive flavor of both our algorithm and lower bound contrasts with much previous work in distributional testing and learning. Copyright © 2019, The Authors. All rights reserved.

关键词： adaptive algorithms

Comments on "A new computing approach for power signal modeling using fractional adaptive algorithms"

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Wahab, Abdul Khan, Shujaat Khan, Farrukh Zeeshan Sector H-12 Islamabad44000 Pakistan Bio-Imaging Signal Processing and Learning Lab. Department of Bio and Brain Engineering Korea Advanced Institute of Science and Technology 291 Daehak-ro Yuseong-gu Daejeon34141 Korea Republic of Department of Computer Science University of Engineering and Technology Taxila Taxila47080 Pakistan

In this paper, some points to the convergence analysis performed in the paper [A new computing approach for power signal modeling using fractional adaptive algorithms, ISA Transactions 68 (2017) 189-202] are presented. It is highlighted that the way the authors prove convergence, suffers lack of correct and valid mathematical justifications. Copyright © 2020, The Authors. All rights reserved.

关键词： adaptive algorithms

ℓ1/ℓ0 Regularized Conjugate Gradient Based Sparse adaptive algorithms

学校读者我要写书评

暂无评论

ℓ1/ℓ0 Regularized Conjugate Gradient Based Sparse Adaptive...

International Conference on Signal Processing and Communications, SPCOM

作者： Rajib Lochan Das Dhirubhai Ambani Institute of Information and Communication Technology Gandhinagar INDIA

ISBN: (数字)9781728188959

ISBN: (纸本)9781728188966

adaptive algorithms in general yield slow convergence rate while identifying systems with colored input. In this context, the adaptive Conjugate Gradient (ACG) algorithm shows fast convergence for colored input. However, the ACG algorithm do not exploit system sparsity. In this paper, the conjugate gradient based sparse adaptive algorithms are proposed. In particular, ℓ 1 and ℓ 0 norm penalties are added to the cost function of the ACG algorithm in order to attract the inactive taps to their optimum (i.e., zero) levels, and the resulting algorithms yield better steady-state performance. Simulation results show that the proposed algorithm outperforms recently proposed ℓ 0 -Recursive Least Square (ℓ 0 -RLS) algorithm.

关键词： Approximation algorithms Steady-state Convergence adaptive algorithms Cost function Simulation adaptive systems

Universal and data-adaptive algorithms for model selection in linear contextual bandits

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Muthukumar, Vidya Krishnamurthy, Akshay Electrical and Computer Engineering Industrial and Systems Engineering Georgia Institute of Technology United States Microsoft Research New York City United States

Model selection in contextual bandits is an important complementary problem to regret minimization with respect to a fixed model class. We consider the simplest non-trivial instance of model-selection: distinguishing a simple multi-armed bandit problem from a linear contextual bandit problem. Even in this instance, current state-of-the-art methods explore in a suboptimal manner and require strong "feature-diversity" conditions. In this paper, we introduce new algorithms that a) explore in a data-adaptive manner, and b) provide model selection guarantees of the form O(dαT1−α) with no feature diversity conditions whatsoever, where d denotes the dimension of the linear model and T denotes the total number of rounds. The first algorithm enjoys a "best-of-both-worlds" property, recovering two prior results that hold under distinct distributional assumptions, simultaneously. The second removes distributional assumptions altogether, expanding the scope for tractable model selection. Our approach extends to model selection among nested linear contextual bandits under some additional assumptions. Copyright © 2021, The Authors. All rights reserved.

关键词： adaptive algorithms

Assessment of adaptive algorithms for automatic control of train braking system

学校读者我要写书评

暂无评论

Assessment of adaptive algorithms for automatic control of t...

International Scientific Conference on Power and Electrical Engineering of Riga Technical University (RTUCON)

作者： Andrejs Potapovs Anatoly Levchenkov Mikhail Gorobetz Rigas Tehniska universitate Riga LV

This paper provides a new insight into the smooth and precise adaptive railway transport braking system development. The system contains a controller with a control program based on an adaptive control algorithm and a current train braking control system ensures an automatic smooth and precise braking of a train and another controller ensures an automatic stopping of the train before the red light. Some of the adaptive search algorithms are studied and the task is to test and select the most suitable and the most effective of them. The computer model and simulation results are described in this paper.

关键词： Rail transportation Software algorithms Algorithm design and analysis adaptive algorithms adaptive systems Control systems Educational institutions

A general system of differential equations to model first order adaptive algorithms

学校读者我要写书评

暂无评论

arXiv 2018年

作者： Silva, André Belotto Da Gazeau, Maxime Université Aix-Marseille 39 rue F. Joliot Curie Marseille13013 France Borealis Ai MaRS Heritage Building 101 College St Suite 350 TorontoONM5G 1L7 Canada

First order optimization algorithms play a major role in large scale machine learning. A new class of methods, called adaptive algorithms, were recently introduced to adjust iteratively the learning rate for each coordinate. Despite great practical success in deep learning, their behaviour and performance on more general loss functions are not well understood. In this paper, we derive a non-autonomous system of differential equations, which is the continuous time limit of adaptive optimization methods. We study the convergence of its trajectories and give conditions under which the differential system, underlying all adaptive algorithms, is suitable for optimization. We discuss convergence to a critical point in the non-convex case and give conditions for the dynamics to avoid saddle points and local maxima. For convex loss function, we introduce a suitable Lyapunov functional which allow us to study its rate of convergence. Several other properties of both the continuous and discrete systems are briey discussed. The differential system studied in the paper is general enough to encompass many other classical algorithms (such as Heavy ball and Nesterov's accelerated method) and allow us to recover several known results for these algorithms. Copyright © 2018, The Authors. All rights reserved.

关键词： adaptive algorithms