Self-adjusting Parameter Research Articles

Investigating individual, loudness-dependent equalization preferences in different driving sound conditions

In automotive audio playback systems, dynamically increasing driving sounds are typically taken into account by applying a generic, i.e., non-individualized, increase in overall level and low-frequency amplification to compensate increased masking. This study investigated the degree of individuality regarding the preferences of noise-dependent level and equalizer settings. A user study with 18 normal-hearing participants was conducted in which individually preferred level-dependent and frequency-dependent amplification parameters were determined using a music-based procedure in quiet and in nine different driving noise conditions. The comparison of self-adjusted parameters suggested that, on average, participants adjusted higher overall levels and more low-frequency amplification in noise than in quiet. However, preferred self-adjusted levels differedmarkedly between participants for the same listening conditions but were quite similar in a re-test session for each participant, indicating that individual preferences were stable and could be reproducibly measured with the employed personalization scheme. Furthermore, the impact of driving noise on individually preferred settings revealed strong interindividual differences, indicating that listeners can differ widely with respect to their individual optimum of how equalizer and level settings should be dynamically adapted to changes in driving conditions.

Self-adjusting offspring population sizes outperform fixed parameters on the cliff function

In the discrete domain, self-adjusting parameters of evolutionary algorithms (EAs) have emerged as a fruitful research area with many runtime analyses showing that self-adjusting parameters can outperform the best fixed parameters. Most existing runtime analyses focus on elitist EAs on simple problems, for which moderate performance gains were shown. Here we consider a much more challenging scenario: the multimodal function Cliff, defined as an example where a (1,λ) EA is effective, and for which the best known upper runtime bound for standard EAs is O(n25).We prove that a (1,λ) EA self-adjusting the offspring population size λ using success-based rules optimises Cliff in O(n) expected generations and O(nlog⁡n) expected evaluations. Along the way, we prove tight upper and lower bounds on the runtime for fixed λ (up to a logarithmic factor) and identify the runtime for the best fixed λ as nη for η≈3.97677 (up to sub-polynomial factors). Hence, the self-adjusting (1,λ) EA outperforms the best fixed parameter by a factor of at least n2.9767.

Self-adjusting Population Sizes for Non-elitist Evolutionary Algorithms: Why Success Rates Matter

Evolutionary algorithms (EAs) are general-purpose optimisers that come with several parameters like the sizes of parent and offspring populations or the mutation rate. It is well known that the performance of EAs may depend drastically on these parameters. Recent theoretical studies have shown that self-adjusting parameter control mechanisms that tune parameters during the algorithm run can provably outperform the best static parameters in EAs on discrete problems. However, the majority of these studies concerned elitist EAs and we do not have a clear answer on whether the same mechanisms can be applied for non-elitist EAs. We study one of the best-known parameter control mechanisms, the one-fifth success rule, to control the offspring population size λ\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\lambda $$\\end{document} in the non-elitist (1,λ)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${(1,\\lambda )}$$\\end{document} EA. It is known that the (1,λ)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${(1,\\lambda )}$$\\end{document} EA has a sharp threshold with respect to the choice of λ\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\lambda $$\\end{document} where the expected runtime on the benchmark function OneMax changes from polynomial to exponential time. Hence, it is not clear whether parameter control mechanisms are able to find and maintain suitable values of λ\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\lambda $$\\end{document}. For OneMax we show that the answer crucially depends on the success rate s (i. e. a one-(s+1)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$(s+1)$$\\end{document}-th success rule). We prove that, if the success rate is appropriately small, the self-adjusting (1,λ)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${(1,\\lambda )}$$\\end{document} EA optimises OneMax in O(n) expected generations and O(nlogn)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$O(n \\log n)$$\\end{document} expected evaluations, the best possible runtime for any unary unbiased black-box algorithm. A small success rate is crucial: we also show that if the success rate is too large, the algorithm has an exponential runtime on OneMax and other functions with similar characteristics.

One Method for Preprocessing Pulse Signal Based on Deep Learning

The collection of pulse signals is accompanied by considerable noise interference, and it is necessary to denoise the collected signals to eliminate the error brought about by the outside world and the instrument itself to the actual data to the greatest extent. Considering this, this article proposes a preprocessing scheme for noise reduction. Firstly, the saturation detection algorithm in signals is based on the gradient, and extreme is utilized to remove the saturation interference. On this basis, the artifact detection module based on complex network connectivity is proposed. Finally, the self-adjusting parameter integer coefficient filtering is utilized to include the baseline drift. The noise inside is filtered out. The experimental results demonstrate that the proposed method, in the case of a similar signal-to-noise ratio, has a mean square error of 15.7 and a shorter convergence time of 0.02s.

Bonobo optimizer (BO): an intelligent heuristic with self-adjusting parameters over continuous spaces and its applications to engineering problems

Bonobo optimizer (BO): an intelligent heuristic with self-adjusting parameters over continuous spaces and its applications to engineering problems

Load Balancing Algorithm of Controller Based on SDN Architecture Under Machine Learning

Abstract With the rapid development of cloud computing and other related services, higher requirements are put forward for network transmission and delay. Due to the inherent distributed characteristics of traditional networks, machine learning technology is difficult to be applied and deployed in network control. The emergence of SDN technology provides new opportunities and challenges for the application of machine learning technology in network management. A load balancing algorithm of Internet of things controller based on data center SDN architecture is proposed. The Bayesian network is used to predict the degree of load congestion, combining reinforcement learning algorithm to make optimal action decision, self-adjusting parameter weight to adjust the controller load congestion, to achieve load balance, improve network security and stability.

ADMM based distributed state estimation for integrated energy system

In order to have an accurate knowledge of systemwide operation states, it is necessary to perform state estimation for the integrated energy system (IES) as the basis of energy management and control. Centralized state estimation is practically infeasible for IES due to the unreliability of communication, the barrier on privacy, and the large scale of integrated systems. This paper proposes a distributed state estimation algorithm based on the alternating direction method of multipliers (ADMM) for IES containing electricity, heat, and natural gas. Various coupling units are taken into full consideration in modeling of IES state estimation to reflect the harmonization of multi energy. On the basis of bilinear measurement model, the state estimation considering nonlinear measurements can be replaced by an equivalent three-stage problem containing two linear state estimations and an intermediate transformation to avoid nonconvex optimization. The three-stage procedure for IES state estimation can be further decoupled over three sub-systems with coordination on coupling units, yielding a fully distributed scheme based on ADMM. A modified ADMM with the self-adjusting penalty parameter is also adopted to enhance the convergence. Simulation results demonstrate the validity and superiority of the proposed algorithm.

Active Disturbance Rejection Control With Self-Adjusting Parameters for Vibration Displacement System of Continuous Casting Mold

Continuous casting mold (CCM) is molding equipment of the casting blank and widely used in iron and steel metallurgy. This paper focuses on the tracking control of vibration displacement of CCM is driven by a servo motor. To address the non-unique inverse solution of the nonlinear periodic function, a new nonlinear processing algorithm is developed. Based on the algorithm, an active disturbance rejection controller (ADRC) with self-adjusting parameters to estimate and compensate the disturbances is proposed for suppressing the disturbances and improving the tracking performance. The simulation results demonstrate the effectiveness of the proposed control scheme.

Multi-strategy ensemble grey wolf optimizer and its application to feature selection

To overcome the limitation of single search strategy of grey wolf optimizer (GWO) in solving various function optimization problems, we propose a multi-strategy ensemble GWO (MEGWO) in this paper. The proposed MEGWO incorporates three different search strategies to update the solutions. Firstly, the enhanced global-best lead strategy can improve the local search ability of GWO by fully exploiting the search space around the current best solution. Secondly, the adaptable cooperative strategy embeds one-dimensional update operation into the framework of GWO to provide a higher population diversity and promote the global search ability. Thirdly, the disperse foraging strategy forces a part of search agents to explore a promising area based on a self-adjusting parameter, which contributes to the balance between the exploitation and exploration. We conducted numerical experiments based on various functions form CEC2014. The obtained results are compared with other three modified GWO and seven state-of-the-art algorithms. Furthermore, feature selection is employed to investigate the effectiveness of MEGWO on real-world applications. The experimental results show that the proposed algorithm which integrate multiple improved search strategies, outperforms other variants of GWO and other algorithms in terms of accuracy and convergence speed. It is validated that MEGWO is an efficient and reliable algorithm not only for optimization of functions with different characteristics but also for real-world optimization problems.

Real-time classifier based on adaptive competitive self-organizing algorithm

This article introduces a novel adaptive competitive self-organizing (ACS) model, with applicability for real-time clustering and vector quantization. An important feature of this model is its dynamic structure and self-adjusting parameters that also offers a solution to the problem of parasitic limit points and consequently in more accurate label assignments. This unsupervised classifier is free of any external control mechanism. Its self-organizing (SO) dynamic is governed by the gradient descent (GD) theory in cooperation with a competition mechanism based on Lotka–Volterra competitive exclusion. The core algorithm of this classifier is based on developing an energy function, where its minima or equilibrium points correspond to the centroid of similar input patterns. Since this energy function is a form of Lyapunov function, it guarantees stabilization of the dynamical trajectories of labels in finite numbers of isolated equilibrium points. This energy function along with other control parameter functions, then, will be the base for the set of ordinary differential equations (ODEs) describing the overall dynamic of our system. Finally, the effectiveness of the proposed ACS model is demonstrated by implementing it on both real and artificial data sets as well as comparing with other well-known clustering methods. ACS method showed a better clustering performance in some categories and an overall comparable rendition.

Self-Adjusted Amplification Parameters Produce Large Between-Subject Variability and Preserve Speech Intelligibility.

The current study used the self-fitting algorithm to allow listeners to self-adjust hearing-aid gain or compression parameters to select gain for speech understanding in a variety of quiet and noise conditions. Thirty listeners with mild to moderate sensorineural hearing loss adjusted gain parameters in quiet and in several types of noise. Outcomes from self-adjusted gain and audiologist-fit gain indicated consistent within-subject performance but a great deal of between-subject variability. Gain selection did not strongly affect intelligibility within the range of signal-to-noise ratios tested. Implications from the findings are that individual listeners have consistent preferences for gain and may prefer gain configurations that differ greatly from National Acoustic Laboratories-based prescriptions in quiet and in noise.

Intelligent tuning method of PID parameters based on iterative learning control for atomic force microscopy

Proportional-integral-derivative (PID) parameters play a vital role in the imaging process of an atomic force microscope (AFM). Traditional parameter tuning methods require a lot of manpower and it is difficult to set PID parameters in unattended working environments. In this manuscript, an intelligent tuning method of PID parameters based on iterative learning control is proposed to self-adjust PID parameters of the AFM according to the sample topography. This method gets enough information about the output signals of PID controller and tracking error, which will be used to calculate the proper PID parameters, by repeated line scanning until convergence before normal scanning to learn the topography. Subsequently, the appropriate PID parameters are obtained by fitting method and then applied to the normal scanning process. The feasibility of the method is demonstrated by the convergence analysis. Simulations and experimental results indicate that the proposed method can intelligently tune PID parameters of the AFM for imaging different topographies and thus achieve good tracking performance.

Self-adjusting parameter control for surrogate-assisted constrained optimization under limited budgets

Constrained optimization of high-dimensional numerical problems plays an important role in many scientific and industrial applications. Function evaluations in many industrial applications are severely limited and often only little analytical information about objective function and constraint functions is available. For such expensive black-box optimization tasks, the constraint optimization algorithm COBRA (Constrained Optimization By Radial Basis Function Approximation) was proposed, making use of RBF (radial basis function) surrogate modeling for both objective and constraint functions. COBRA has shown remarkable success in solving reliably complex benchmark problems in less than 500 function evaluations. Unfortunately, COBRA requires careful adjustment of parameters in order to do so.In this work we present a new algorithm SACOBRA (Self-Adjusting COBRA), which is based on COBRA and capable of achieving high-quality results with very few function evaluations and no parameter tuning. It is shown with the help of performance profiles on a set of benchmark problems (G-problems, MOPTA08) that SACOBRA consistently outperforms COBRA algorithms with different fixed parameter settings. We analyze the importance of the new elements in SACOBRA and show that each element of SACOBRA plays a role to boost up the overall optimization performance. We discuss the reasons and get in this way a better understanding of high-quality RBF surrogate modeling.

Optimal Static and Self-Adjusting Parameter Choices for the $$(1+(\lambda ,\lambda ))$$ ( 1 + ( λ , λ ) ) Genetic Algorithm

The $$(1+(\lambda ,\lambda ))$$ genetic algorithm proposed in Doerr et al. (Theor Comput Sci 567:87–104, 2015) is one of the few examples for which a super-constant speed-up of the expected optimization time through the use of crossover could be rigorously demonstrated. It was proven that the expected optimization time of this algorithm on OneMax is $$O(\max \{n \log (n) / \lambda , \lambda n\})$$ for any offspring population size $$\lambda \in \{1,\ldots ,n\}$$ (and the other parameters suitably dependent on $$\lambda $$ ) and it was shown experimentally that a self-adjusting choice of $$\lambda $$ leads to a better, most likely linear, runtime. In this work, we study more precisely how the optimization time depends on the parameter choices, leading to the following results on how to optimally choose the population size, the mutation probability, and the crossover bias both in a static and a dynamic fashion. For the mutation probability and the crossover bias depending on $$\lambda $$ as in Doerr et al. (2015), we improve the previous runtime bound to $$O(\max \{n \log (n) / \lambda , n \lambda \log \log (\lambda )/\log (\lambda )\})$$ . This expression is minimized by a value of $$\lambda $$ slightly larger than what the previous result suggested and gives an expected optimization time of $$O\left( n \sqrt{\log (n) \log \log \log (n) / \log \log (n)}\right) $$ . We show that no static choice in the three-dimensional parameter space of offspring population, mutation probability, and crossover bias gives an asymptotically better runtime. We also prove that the self-adjusting parameter choice suggested in Doerr et al. (2015) outperforms all static choices and yields the conjectured linear expected runtime. This is asymptotically optimal among all possible parameter choices.

Automation and Combination of Linear-Programming Based Stabilization Techniques in Column Generation

The convergence of a column generation algorithm can be improved in practice by using stabilization techniques. Smoothing and proximal methods based on penalizing the deviation from the incumbent dual solution have become standards of the domain. Interpreting column generation as cutting plane strategies in the dual problem, we analyze the mechanisms on which stabilization relies. In particular, the link is established between smoothing and in-out separation strategies to derive generic convergence properties. For penalty function methods as well as for smoothing, we describe proposals for parameter self-adjusting schemes. Such schemes make initial parameter tuning less of an issue as corrections are made dynamically. Such adjustments also allow to adapt the parameters to the phase of the algorithm. We provide extensive test reports that validate our self-adjusting parameter scheme and highlight their performances. Our results also show that using smoothing in combination with penalty function yields a cumulative effect on convergence speed-ups.

Fast Unsupervised Bayesian Image Segmentation With Adaptive Spatial Regularisation.

This paper presents a new Bayesian estimation technique for hidden Potts-Markov random fields with unknown regularisation parameters, with application to fast unsupervised K -class image segmentation. The technique is derived by first removing the regularisation parameter from the Bayesian model by marginalisation, followed by a small-variance-asymptotic (SVA) analysis in which the spatial regularisation and the integer-constrained terms of the Potts model are decoupled. The evaluation of this SVA Bayesian estimator is then relaxed into a problem that can be computed efficiently by iteratively solving a convex total-variation denoising problem and a least-squares clustering ( K -means) problem, both of which can be solved straightforwardly, even in high-dimensions, and with parallel computing techniques. This leads to a fast fully unsupervised Bayesian image segmentation methodology in which the strength of the spatial regularisation is adapted automatically to the observed image during the inference procedure, and that can be easily applied in large 2D and 3D scenarios or in applications requiring low computing times. Experimental results on synthetic and real images, as well as extensive comparisons with state-of-the-art algorithms, confirm that the proposed methodology offer extremely fast convergence and produces accurate segmentation results, with the important additional advantage of self-adjusting regularisation parameters.

GATK hard filtering: tunable parameters to improve variant calling for next generation sequencing targeted gene panel data

BackgroundNGS technology represents a powerful alternative to the standard Sanger sequencing in the context of clinical setting. The proprietary software that are generally used for variant calling often depend on preset parameters that may not fit in a satisfactory manner for different genes.GATK, which is widely used in the academic world, is rich in parameters for variant calling. However the self-adjusting parameter calibration of GATK requires data from a large number of exomes. When these are not available, which is the standard condition of a diagnostic laboratory, the parameters must be set by the operator (hard filtering). The aim of the present paper was to set up a procedure to assess the best parameters to be used in the hard filtering of GATK. This was pursued by using classification trees on true and false variants from simulated sequences of a real dataset data.ResultsWe simulated two datasets, with different coverages, including all the sequence alterations identified in a real dataset according to their observed frequencies. Simulated sequences were aligned with standard protocols and then regression trees were built up to identify the most reliable parameters and cutoff values to discriminate true and false variant calls. Moreover, we analyzed flanking sequences of region presenting a high rate of false positive calls observing that such sequences present a low complexity make up.ConclusionsOur results showed that GATK hard filtering parameter values can be tailored through a simulation study based-on the DNA region of interest to ameliorate the accuracy of the variant calling.

Channel self-adjusting filtered-x LMS algorithm for active control of vehicle road noise

Current active road noise control (ARNC) systems, configured with the standard filtered-x least mean square (FxLMS) algorithm, are not sufficient enough to yield an ideal noise reduction over a broad frequency range. This is because ARNC systems generally employ multiple reference signals, which has an inherent limitation of the channel-dependent convergence behaviour due to dynamic characteristics amongst reference signals. In this study, an effective ARNC system with the channel self-adjusting FxLMS (CSFxLMS) algorithm by incorporating the self-adjusting parameter on each reference signal path is proposed, which is to minimise the effect of its dynamics characteristics in different reference channels. To validate the effectiveness of the proposed algorithm, numerical simulations using measured road noise response is conducted. Results show that the performance of the proposed CSFxLMS algorithm is significantly better as compared to the conventional FxLMS algorithm, and is able to achieve around 5 dBA reductions at the driver's ear position.

Individual differences in self-adjustment of hearing aid amplification

Listening in noisy environments remains a common complaint among hearing aid users. Self-adjustment of amplification parameters could improve listener satisfaction in noise. Previous data (Nelson et al., AAS 2015) indicated considerable variability among hearing-impaired listeners when self-adjusting amplification in noisy environments, but the sources of variability are not understood. Data will be presented from 30 adult listeners with mild to moderate hearing loss who self-adjusted amplification parameters in laboratory-simulated restaurant environments. Participants listened using a real-time simulation of a multichannel compression hearing aid and adjusted gain/compression parameters via a simple user interface (EarMachine). On average, listeners selected less gain relative to NAL-NL2 targets, and they reduced gain as noise levels increased. Hearing loss did not predict variation in self-selected amplification. Small but significant gender differences were observed. Female subjects selected more high-frequency gain (2000–8000 Hz) but less low-frequency gain than male subjects. Listeners with prior hearing-aid experience selected more high-frequency gain than listeners with no hearing-aid experience. Younger age was associated with selecting more low-frequency gain. Within-group variability was large compared to mean differences between groups and only a small proportion of variance was explained. These data suggest that self-adjustment of amplification was not easily predicted by listener-specific factors.

Dynamic analysis and control strategy of wet clutches during torque phase of gear shift

Control over the clutch during gear shift is the key technique in automatic transmission shift control and this control largely decides the shift quality. The engagement of the on-coming clutch and the disengagement of the off-going clutch should be controlled synchronously and accurately to achieve a smooth shift. Optimal control of clutch torque exchange phase is required to achieve this synchronous and accurate control. Shift quality can be easily influenced with inappropriate control parameters because of the many factors that affect the torque phase, as well as the use of closed-loop control. In this study, a dynamic model of frictional clutch with electro-hydraulic shift control is built. On the basis of the study of the influence of clutch-related parameters, an optimal control model is proposed with the closed-loop control together with an adaptive control method based on micro-slip clutch control. This control method can continually self-adjust control parameters to prevent shock, improve control precision and effectively reduce the calibration work after. Test result shows that this closed loop with adaptive method ensures consistently good shift quality.