Automatic Parameter Selection Research Articles

Pole Identification using discrete Laguerre expansion and variable projection

We propose a novel algorithm for identifying the poles of transfer functions describing SISO-LTI (single input single output, linear time invariant) systems. Our Identification method works in the frequency domain and consists of two parts. In the first part, we extend a discrete Laguerre expansion based method with an automatic parameter selection scheme. This allows us to find an initial estimate of the poles of SISO-LTI transfer functions without the need for human intuition. Then, in the second part, we propose a novel optimization problem to improve our initial estimates. The proposed optimization aims to reduce the least squared error of a parameterized model, which can be interpreted as an orthogonal projection of the system's frequency response onto a subspace spanned by Generalized Orthogonal Rational Basis functions (GOBFs). We solve the corresponding nonlinear optimization task using gradient based methods, where we can analytically calculate the gradient of the error functional. Through robust numerical experiments, we investigate the behavior of the developed methods and show that they work even in scenarios, when the transfer function has a high number of poles.

Coarse Point Cloud Registration Based on Variational Functionals

Point cloud collection forming a 3D scene typically uses information from multiple data scans. The common approach is to register the point cloud pairs consequentially using a variant of the iterative closest point (ICP) algorithm, but most versions of the ICP algorithm only work correctly for a small movement between two point clouds. This makes it difficult to accumulate multiple scans. Global registration algorithms are also known, which theoretically process point clouds at arbitrary initial positions. Recently, a multiparameter variational functional was described and used in the ICP variant to register point clouds at arbitrary initial positions. The disadvantage of this algorithm was the need for manual selection of parameters. In this paper, a modified version of the algorithm with automatic selection of the model parameters is proposed. The proposed algorithm is a fusion of the ICP and RANSAC concepts. Moreover, the algorithm can be parallelized. The performance of the proposed algorithm is compared with that of known global registration algorithms.

Adaptive Filtering for the Maternal Respiration Signal Attenuation in the Uterine Electromyogram.

The electrohysterogram (EHG) is the uterine muscle electromyogram recorded at the abdominal surface of pregnant or non-pregnant woman. The maternal respiration electromyographic signal (MR-EMG) is one of the most relevant interferences present in an EHG. Alvarez (Alv) waves are components of the EHG that have been indicated as having the potential for preterm and term birth prediction. The MR-EMG component in the EHG represents an issue, regarding Alv wave application for pregnancy monitoring, for instance, in preterm birth prediction, a subject of great research interest. Therefore, the Alv waves denoising method should be designed to include the interference MR-EMG attenuation, without compromising the original waves. Adaptive filter properties make them suitable for this task. However, selecting the optimal adaptive filter and its parameters is an important task for the success of the filtering operation. In this work, an algorithm is presented for the automatic adaptive filter and parameter selection using synthetic data. The filter selection pool comprised sixteen candidates, from which, the Wiener, recursive least squares (RLS), householder recursive least squares (HRLS), and QR-decomposition recursive least squares (QRD-RLS) were the best performers. The optimized parameters were L = 2 (filter length) for all of them and λ = 1 (forgetting factor) for the last three. The developed optimization algorithm may be of interest to other applications. The optimized filters were applied to real data. The result was the attenuation of the MR-EMG in Alv waves power. For the Wiener filter, power reductions for quartile 1, median, and quartile 3 were found to be −16.74%, −20.32%, and −15.78%, respectively (p-value = 1.31 × 10−12).

A sparse additive model for high-dimensional interactions with an exposure variable

A conceptual paradigm for onset of a new disease is often considered to be the result of changes in entire biological networks whose states are affected by a complex interaction of genetic and environmental factors. However, when modeling a relevant phenotype as a function of high dimensional measurements, power to estimate interactions is low, the number of possible interactions could be enormous and their effects may be non-linear. A method called sail for detecting non-linear interactions with a key environmental or exposure variable in high-dimensional settings which respects the strong or weak heredity constraints is proposed. It is proven that asymptotically, sail possesses the oracle property, i.e., it performs as well as if the true model were known in advance. A computationally efficient fitting algorithm with automatic tuning parameter selection, which scales to high-dimensional datasets is proposed. Simulation results show that sail outperforms existing penalized regression methods in terms of prediction accuracy and support recovery when there are non-linear interactions with an exposure variable. sail is applied to detect non-linear interactions between genes and a prenatal psychosocial intervention program on cognitive performance in children at 4 years of age. Results show that individuals who are genetically predisposed to lower educational attainment are those who stand to benefit the most from the intervention. The proposed algorithms are implemented in an R package available on CRAN (https://cran.r-project.org/package=sail).

Prediction of Economic Operation Index Based on Support Vector Machine

Economic forecasting is not only an important field of economic research but also attracts extensive public attention. The results of the study are directly related to the accurate understanding and view of the economic situation, which in turn affects the rational formulation of macroeconomic policies. However, traditional estimation methods are often limited by expert experience and simple mathematical models, which are difficult to deal with on nonlinear models and do not meet the objective requirements for predicting macroeconomic performance indicators. Support vector machine is a popular new data mining technique. Benefiting from its good theoretical foundation and good generalization performance, SVM has become the starting point of research in recent years. Combining support vector machine research, fuzzy theory, and macroeconomic performance estimation, it attempts to develop a method for predicting macroeconomic function indicators based on support vector machines and expand the theory and projects of support vector machines. In addition, empirical evaluations of early financial forecasts are conducted to integrate theoretical and practical data. The characteristics of the traditional evaluation system are discussed in detail, and the standard model of the prediction system is established, including classical and modern forecasting theories and their forecasting systems. The theory of statistical learning and the theory and basic features of support vector machines are also discussed. For the analysis of the internal relationship between the distribution pattern, SVM, and the forecast of macroeconomic performance indicators, predictable economic activity forecast can be considered as a distribution system. Combining SVM with economic forecasting, SWE intelligent economic operation index forecasting is used for the first time, the automatic selection of symbolic parameters is recognized, and some algorithms and in-depth analysis methods are provided. The experimental results show that the prediction of economic operation indicators based on support vector machine technology is 57% faster than the traditional prediction method and the prediction accuracy rate is as high as 98.56%.

Automatic parameter selection for electron ptychography via Bayesian optimization

Electron ptychography provides new opportunities to resolve atomic structures with deep sub-angstrom spatial resolution and to study electron-beam sensitive materials with high dose efficiency. In practice, obtaining accurate ptychography images requires simultaneously optimizing multiple parameters that are often selected based on trial-and-error, resulting in low-throughput experiments and preventing wider adoption. Here, we develop an automatic parameter selection framework to circumvent this problem using Bayesian optimization with Gaussian processes. With minimal prior knowledge, the workflow efficiently produces ptychographic reconstructions that are superior to those processed by experienced experts. The method also facilitates better experimental designs by exploring optimized experimental parameters from simulated data.

ADMM-Based Residual Whiteness Principle for Automatic Parameter Selection in Single Image Super-Resolution Problems

We propose an automatic parameter selection strategy for the single image super-resolution problem for images corrupted by blur and additive white Gaussian noise with unknown standard deviation. The proposed approach exploits the structure of both the down-sampling and the blur operators in the frequency domain and computes the optimal regularisation parameter as the one optimising a suitably defined residual whiteness measure. Computationally, the proposed strategy relies on the fast solution of generalised Tikhonov ell _2–ell _2 problems as proposed in Zhao et al. (IEEE Trans Image Process 25:3683–3697, 2016). These problems naturally appear as substeps of the Alternating Direction Method of Multipliers used to solve single image super-resolution problems with non-quadratic, non-smooth, sparsity-promoting regularisers both in convex and in non-convex regimes. After detailing the theoretical properties allowing to express the whiteness functional in a compact way, we report an exhaustive list of numerical experiments proving the effectiveness of the proposed approach for different type of problems, in comparison with well-known parameter selection strategies such as, e.g., the discrepancy principle.

Copula link-based additive models for bivariate time-to-event outcomes with general censoring scheme

Bivariate survival outcomes arise frequently in applied studies where the occurrence of two events of interest are associated. Often the exact event times are unknown due to censoring which can manifest in various forms. A general and flexible copula regression model that can handle bivariate survival data subject to various censoring mechanisms, which include a mixture of uncensored, left-, right-, and interval-censored data, is proposed. The proposal permits to specify all model parameters as flexible functions of covariate effects, flexibly model the baseline survival functions by means of monotonic P-splines, characterise the marginals via transformations of the survival functions which yield, e.g., the proportional hazards and odds models as special cases, and model the dependence between events using a wide variety of copulae. The algorithm is based on a computationally efficient and stable penalised maximum likelihood estimation approach with integrated automatic multiple smoothing parameter selection. The proposed model is evaluated in a simulation study and illustrated using data from the Age-Related Eye Disease Study. The modelling framework has been incorporated in the newly-revised R package GJRM, hence allowing any user to fit the desired model(s) and produce easy-to-interpret numerical and visual summaries.

Detection of hierarchical crowd activity structures in geographic point data.

The pervasive adoption of GPS-enabled sensors has lead to an explosion on the amount of geolocated data that captures a wide range of social interactions. Part of this data can be conceptualized as event data, characterized by a single point signal at a given location and time. Event data has been used for several purposes such as anomaly detection and land use extraction, among others. To unlock the potential offered by the granularity of this new sources of data it is necessary to develop new analytical tools stemming from the intersection of computational science and geographical analysis. Our approach is to link the geographical concept of hierarchical scale structures with density based clustering in databases with noise to establish a common framework for the detection of crowd activity hierarchical structures in geographic point data. Our contribution is threefold: first, we develop a tool to generate synthetic data according to a distribution commonly found on geographic event data sets; second, we propose an improvement of the available methods for automatic parameter selection in density-based spatial clustering of applications with noise (DBSCAN) algorithm that allows its iterative application to uncover hierarchical scale structures on event databases and, lastly, we propose a framework for the evaluation of different algorithms to extract hierarchical scale structures. Our results show that our approach is successful both as a general framework for the comparison of crowd activity detection algorithms and, in the case of our automatic DBSCAN parameter selection algorithm, as a novel approach to uncover hierarchical structures in geographic point data sets.

Two-Dimensional Gravity Inversion of Basement Relief for Geothermal Energy Potentials at the Harrat Rahat Volcanic Field, Saudi Arabia, Using Particle Swarm Optimization

We invert gravity and magnetic anomalies for basement relief at the Harrat Rahat Volcanic Field (HRVF) for the purpose of evaluating its geothermal energy prospects. HRVF is dominated by basaltic scoria cones and other volcanic rocks overlying the Proterozoic basement. The area considered for this study is located within the northern HRVF and consists mainly of alkali basalts with lesser amounts of benmoreite, mugearite, hawaiite, and trachyte. Our approach adopts a global optimization technique using Particle Swarm Optimization with automated parameter selection, and a two-dimensional gravity-magnetic (GM) forward modeling procedure. The results of the PSO-based approach indicate a depth to the basement at 0.10–624 m, with greater depths within the central region of a solitary anomalous density body in the HRVF. The obtained basement geometry is corroborated by the depth estimates obtained from other potential field inversion methods. The regions with higher prospects are mapped for a targeted future geothermal energy exploration at the HRVF, based on our inversion results.

Flexible parameter selection methods for Rician noise removal with convergence guarantee

Restoring images corrupted by Rician noise is a challenging issue in the field of medical image processing. In the existing variational methods, there is a balancing parameter between the regularization term and the fidelity term. However, it is very hard to find the optimal parameter. In this paper, we study the total variation-based Rician noise removal model with spatially varying parameters. We propose flexible and automatic parameter selection strategies to balance the regularization extent between different kinds of image regions. A modified alternating direction method of multipliers is derived to solve the non-convex model efficiently. Theoretically, we prove that if a selection strategy satisfies some reasonable conditions, the convergence of the proposed algorithm is guaranteed. Numerical results demonstrate that the proposed method with automatic parameter selection can better preserve the structures and fine textures than other closely related methods.

Automated Parameter Selection for Accelerated MRI Reconstruction via Low-Rank Modeling of Local k-Space Neighborhoods

PurposeImage quality in accelerated MRI rests on careful selection of various reconstruction parameters. A common yet tedious and error-prone practice is to hand-tune each parameter to attain visually appealing reconstructions. Here, we propose a parameter tuning strategy to automate hybrid parallel imaging (PI) – compressed sensing (CS) reconstructions via low-rank modeling of local k-space neighborhoods (LORAKS) supplemented with sparsity regularization in wavelet and total variation (TV) domains. MethodsFor low-rank regularization, we leverage a soft-thresholding operation based on singular values for matrix rank selection in LORAKS. For sparsity regularization, we employ Stein's unbiased risk estimate criterion to select the wavelet regularization parameter and local standard deviation of reconstructions to select the TV regularization parameter. Comprehensive demonstrations are presented on a numerical brain phantom and in vivo brain and knee acquisitions. Quantitative assessments are performed via PSNR, SSIM and NMSE metrics. ResultsThe proposed hybrid PI-CS method improves reconstruction quality compared to PI-only techniques, and it achieves on par image quality to reconstructions with brute-force optimization of reconstruction parameters. These results are prominent across several different datasets and the range of examined acceleration rates. ConclusionA data-driven parameter tuning strategy to automate hybrid PI-CS reconstructions is presented. The proposed method achieves reliable reconstructions of accelerated multi-coil MRI datasets without the need for exhaustive hand-tuning of reconstruction parameters.

Towards Automatic Parameter Selection for Multifidelity Surrogate-Based Optimization

Towards Automatic Parameter Selection for Multifidelity Surrogate-Based Optimization

RCDPeaks: memory-efficient density peaks clustering of long molecular dynamics.

Density Peaks is a widely spread clustering algorithm that has been previously applied to Molecular Dynamics (MD) simulations. Its conception of cluster centers as elements displaying both a high density of neighbors and a large distance to other elements of high density, particularly fits the nature of a geometrical converged MD simulation. Despite its theoretical convenience, implementations of Density Peaks carry a quadratic memory complexity that only permits the analysis of relatively short trajectories. Here, we describe DP+, an exact novel implementation of Density Peaks that drastically reduces the RAM consumption in comparison to the scarcely available alternatives designed for MD. Based on DP+, we developed RCDPeaks, a refined variant of the original Density Peaks algorithm. Through the use of DP+, RCDPeaks was able to cluster a one-million frames trajectory using less than 4.5 GB of RAM, a task that would have taken more than 2 TB and about 3× more time with the fastest and less memory-hunger alternative currently available. Other key features of RCDPeaks include the automatic selection of parameters, the screening of center candidates and the geometrical refining of returned clusters. The source code and documentation of RCDPeaks are free and publicly available on GitHub (https://github.com/LQCT/RCDPeaks.git). Supplementary data are available at Bioinformatics online.

Розроблення програмного та алгоритмічного забезпечення для прогнозування курсу криптовалют з використанням методів фрактального аналізу

The work created software and algorithmic support for modeling and forecasting the Bitcoin cryptocurrency using the ARFIMA (AutoRegressive Fractionally Integrated Moving Average) fractal model. Time series forecasting models (autoregressive, fractal) were analyzed. The selection of the most appropriate parameters of the selected fractal model was also carried out to maximize accuracy in view of the RMSE metric. The series were analyzed for trend, seasonality, white noise, non-stationarity and long-term memory. The Hurst indicators were studied and the algorithm for choosing the optimal parameter d of fractal differentiation of the ARFIMA model was adapted. The choice of software tools for implementing algorithms and forecasting models using the Python programming language version 3.6.5 using the pandas version 1.1.3 and numpy version 1.19.2 libraries is justified. In order to forecast the time series, the programming language R version 4.1.3 was used, along with the forecast version 8.16 and arfima version 1.8.0 libraries. The software implementation of the ARFIMA fractal model was carried out. Transferred the application to the Google Colab cloud service using Google Drive storage for storing data and forecasting results. The results of comparing the effectiveness of the created fractal model with the same model with automatic selection of parameters, as well as with the most appropriate autoregression model on different sizes of training and test data were obtained. It was established that a larger amount of both training and test data clearly favors fractal models, since in this case there is a long-lasting effect, that is, a pronounced long memory in the second time series. The result is a software system that can be used by investors and ordinary people to analyze and forecast their chosen cryptocurrency using a modern fractal modeling approach. It is important to always check the data and clean up anomalous deviations that cause error in the prediction estimate.

АRIMA-МОДЕЛІ: МОДЕЛЮВАННЯ ТА ПРОГНОЗУВАННЯ ЦІНИ АКЦІЙ

Modeling the dynamics and forecasting of financial market indicators is of interest to its participants and analysts, as well as scientific circles. The implementation of the task of researching market indicators involves the selection of appropriate methods, tools and resources. The popularity and a significant number of approbations belongs to technical analysis, based on the visual analysis of time series using the construction of special charts and graphical models (figures). Technical analysis is subjective, therefore, in addition to it, methods are used that involve the use of a mathematical apparatus. ARIMA model has shown its effectiveness in working with different time series and has become a powerful tool for obtaining accurate forecasts. The algorithm for constructing such a model involves performing a number of mathematical calculations, which can cause difficulties. But thanks to modern software capabilities, for example, the R programming language, statistical analysis of time series, namely the construction of an ARIMA model, is implemented quickly and with the ability to obtain graphical and numerical results. That is why building an ARIMA model using the R language for modeling the price of a company's shares is of practical importance, which will allow you to get a forecast and make decisions during asset purchase and sale transactions in the financial market. In this work, the algorithm for constructing an ARIMA model is implemented using the R-studio environment in 3 stages (with the consistent use of the corresponding library functions) using the example of PepsiCo prices of stocks time series (monthly and daily data). The graphs of the series were constructed, the series were tested for stationarity, an assumption was made about the value of the model parameters according to the preliminary analysis, automatic selection of parameters was also used, and the corresponding models were built. All constructed models were tested for adequacy through appropriate tests and criteria, as well as the quality of approximation of the actual data by the model. Forecast values were obtained, presented graphically in comparison with the actual data of the share price, and the accuracy of the forecast was calculated.

Neural Approximators for Variable-Order Fractional Calculus Operators (VO-FC)

The paper presents research on the approximation of variable-order fractional operators by recurrent neural networks. The research focuses on two basic variable-order fractional operators, i.e., integrator and differentiator. The study includes variations of the order of each fractional operator. The recurrent neural network architecture based on GRU (Gated Recurrent Unit) cells functioned as a neural approximation for selected fractional operators. The paper investigates the impact of the number of neurons in the hidden layer, treated as a hyperparameter, on the quality of modeling error. Training of the established recurrent neural network was performed on synthetic data sets. Data for training was prepared based on the modified Grünwald-Letnikov definition of variable-order fractional operators suitable for convenient numerical computing without memory effects. The research presented in this paper showed that recurrent network architecture based on GRU-type cells can satisfactorily approximate targeted simple yet functional variable-order fractional operators with minor modeling errors. In addition, the research also compares the presented solution with basic and recurrent neural networks that utilize Tapped Delay Lines (TDL) in their structure. The presented solution is a novel approach to the approximation of VO-FC operators. It has the advantage of automatic selection of neural approximator parameters by optimization based on data customized for specific requirements.

Rough Diamond Machining Process Based on Neural Network

In recent years, mechanical engineers have improved the machining accuracy and machining efficiency by different techniques. This paper studied the rough diamond machining process based on neural network. By using the RBF neural network algorithm, in the process of machining, automatic selection of the appropriate machining parameters without human intervention effectively has improved the machining precision and speed which has demonstrable effect.

Automatic Roi Recognition and Parameters Selection for Digital Image Correlation in Measuring Structures with Complex Shapes

Automatic Roi Recognition and Parameters Selection for Digital Image Correlation in Measuring Structures with Complex Shapes

COL0RME: Super-resolution microscopy based on sparse blinking/fluctuating fluorophore localization and intensity estimation.

To overcome the physical barriers caused by light diffraction, super-resolution techniques are often applied in fluorescence microscopy. State-of-the-art approaches require specific and often demanding acquisition conditions to achieve adequate levels of both spatial and temporal resolution. Analyzing the stochastic fluctuations of the fluorescent molecules provides a solution to the aforementioned limitations, as sufficiently high spatio-temporal resolution for live-cell imaging can be achieved using common microscopes and conventional fluorescent dyes. Based on this idea, we present COL0RME, a method for covariance-based super-resolution microscopy with intensity estimation, which achieves good spatio-temporal resolution by solving a sparse optimization problem in the covariance domain and discuss automatic parameter selection strategies. The method is composed of two steps: the former where both the emitters' independence and the sparse distribution of the fluorescent molecules are exploited to provide an accurate localization; the latter where real intensity values are estimated given the computed support. The paper is furnished with several numerical results both on synthetic and real fluorescence microscopy images and several comparisons with state-of-the art approaches are provided. Our results show that COL0RME outperforms competing methods exploiting analogously temporal fluctuations; in particular, it achieves better localization, reduces background artifacts, and avoids fine parameter tuning.