Search

• We propose a class of kernel interpolation-based methods, called kernel reconstruction learning, for solving machine learning problems. • We prove a reconstruction representer theorem, which indicates that conventional kernel methods can be viewed as special cases of kernel reconstruction learning. • We propose the kernel reconstruction vector machine, kernel reconstruction logistic regression, and kernel reconstruction density estimation methods, and show that they outperform popular kernel methods. This paper proposes a class of kernel interpolation-based methods, called kernel reconstruction learning, for solving machine learning problems. Kernel reconstruction learning uses kernel interpolators to reconstruct the unknown functions, which are needed to estimate in the problem, with estimated function values at selected knots. It can be applied to any learning problem that involves function estimation. We prove a reconstruction representer theorem, which indicates that conventional kernel methods, including kernel ridge regression, kernel support vector machine, and kernel logistic regression, can be viewed as special cases of kernel reconstruction learning. Furthermore, kernel reconstruction learning provides new algorithms for large datasets. The kernel reconstruction vector machine, kernel reconstruction logistic regression, and kernel reconstruction density estimation are discussed in detail. With appropriate implementations, they are shown to have higher prediction/estimation accuracy and/or less computational cost than popular kernel methods. [ABSTRACT FROM AUTHOR]

Background and aims: With metabolic dysfunction-associated fatty liver disease (MAFLD) incidence and prevalence sharply increasing globally, there is an urgent need for non-invasive diagnostic tests to accurately screen high-risk MAFLD patients for liver inflammation and fibrosis. We aimed to develop a novel sequential algorithm based on N-terminal propeptide of type 3 collagen (PRO-C3) for disease risk stratification in patients with MAFLD. Methods: A derivation and independent validation cohort of 327 and 142 patients with biopsy-confirmed MAFLD were studied. We compared the diagnostic performances of various non-invasive scores in different disease states, and a novel sequential algorithm was constructed by combining the best performing non-invasive scores. Results: For patients with high-risk progressive steatohepatitis (i.e., steatohepatitis + NAFLD activity score ≥ 4 + F ≥ 2), the AUROC of FAST score was 0.801 (95% confidence interval (CI): 0.739–0.863), and the negative predictive value (NPV) was 0.951. For advanced fibrosis (≥ F3) and cirrhosis (F4), the AUROCs of ADAPT and Agile 4 were 0.879 (95%CI 0.825–0.933) and 0.943 (95%CI 0.892–0.994), and the NPV were 0.972 and 0.992. Sequential algorithm of ADAPT + Agile 4 combination was better than other combinations for risk stratification of patients with severe fibrosis (AUROC = 0.88), with similar results in the validation cohort. Meanwhile, in all subgroup analyses (stratifying by sex, age, diabetes, NAS, BMI and ALT), ADAPT + Agile 4 had a good diagnostic performance. Conclusions: The new sequential algorithm reliably identifies liver inflammation and fibrosis in MAFLD, making it easier to exclude low-risk patients and recommending high-risk MAFLD patients for clinical trials and emerging pharmacotherapies. [ABSTRACT FROM AUTHOR]

We consider the nonparametric estimation of an S‐shaped regression function. The least squares estimator provides a very natural, tuning‐free approach, but results in a non‐convex optimization problem, since the inflection point is unknown. We show that the estimator may nevertheless be regarded as a projection onto a finite union of convex cones, which allows us to propose a mixed primal‐dual bases algorithm for its efficient, sequential computation. After developing a projection framework that demonstrates the consistency and robustness to misspecification of the estimator, our main theoretical results provide sharp oracle inequalities that yield worst‐case and adaptive risk bounds for the estimation of the regression function, as well as a rate of convergence for the estimation of the inflection point. These results reveal not only that the estimator achieves the minimax optimal rate of convergence for both the estimation of the regression function and its inflection point (up to a logarithmic factor in the latter case), but also that it is able to achieve an almost‐parametric rate when the true regression function is piecewise affine with not too many affine pieces. Simulations and a real data application to air pollution modelling also confirm the desirable finite‐sample properties of the estimator, and our algorithm is implemented in the R package Sshaped. [ABSTRACT FROM AUTHOR]

We introduce a modern, optimized, and publicly available implementation of the sequential Information Bottleneck clustering algorithm, which strikes a highly competitive balance between clustering quality and speed. We describe a set of optimizations that make the algorithm computation more efficient, particularly for the common case of sparse data representation. The results are substantiated by an extensive evaluation that compares the algorithm to commonly used alternatives, focusing on the practically important use case of text clustering. The evaluation covers a range of publicly available benchmark datasets and a set of clustering setups employing modern word and sentence embeddings obtained by state-of-the-art neural models. The results show that in spite of using the more basic Term-Frequency representation, the proposed implementation provides a highly attractive trade-off between quality and speed that outperforms the alternatives considered. This new release facilitates the use of the algorithm in real-world applications of text clustering. [ABSTRACT FROM AUTHOR]

The generalized linear models (GLMs) are widely used in statistical analysis and the related design issues are undoubtedly challenging. The state‐of‐the‐art works mostly apply to design criteria on the estimates of regression coefficients. The prediction accuracy is usually critical in modern decision‐making and artificial intelligence applications. It is of importance to study optimal designs from the prediction aspects for GLMs. In this work, we consider Elastic I‐optimality as a prediction‐oriented design criterion for GLMs, and develop an efficient algorithm for such EI‐optimal designs. By investigating theoretical properties for the optimal weights of any set of design points and extending the general equivalence theorem to the EI‐optimality for GLMs, the proposed efficient algorithm adequately combines the Fedorov–Wynn algorithm and the multiplicative algorithm. It achieves great computational efficiency with guaranteed convergence. Numerical examples are conducted to evaluate the feasibility and computational efficiency of the proposed algorithm. [ABSTRACT FROM AUTHOR]

We introduce a modern, optimized, and publicly available implementation of the sequential Information Bottleneck clustering algorithm, which strikes a highly competitive balance between clustering quality and speed. We describe a set of optimizations that make the algorithm computation more efficient, particularly for the common case of sparse data representation. The results are substantiated by an extensive evaluation that compares the algorithm to commonly used alternatives, focusing on the practically important use case of text clustering. The evaluation covers a range of publicly available benchmark datasets and a set of clustering setups employing modern word and sentence embeddings obtained by state-of-the-art neural models. The results show that in spite of using the more basic Term-Frequency representation, the proposed implementation provides a highly attractive trade-off between quality and speed that outperforms the alternatives considered. This new release facilitates the use of the algorithm in real-world applications of text clustering.

Transfer of a polymer across a membrane is the subject of extensive theoretical and experimental works. One of the most fruitful approaches to such studies relies on the usage of Monte Carlo simulations. In those methods, the polymers are viewed as an abstract chain-like structure, and among the principal goals of the simulation-analysis is to relate their passage time through a cell boundary with various characteristics describing both the structures and their environment. In this paper, we develop probabilistic models for a passage time of the chain-like-structures through a thick membrane. We examine the impact of the boundary thickness on the passage time through a flat membrane as well as its dependence on the polymer length. All our experiments are carried out with the help of our own-developed sequential algorithm that incorporates the tension-propagation mechanism. [ABSTRACT FROM AUTHOR]

Voltage source converter high voltage direct current (VSC-HVDC) transmission technology is increasing its share in AC transmission due to important advantages such as independent P and Q control and lower harmonics. Nowadays, this technology is used to connect far-away remote offshore wind-power plants to mainland networks. VSC-HVDC is the main actor in developing the future multi-terminal DC grids and supergrids. This paper proposes a sequential algorithm to calculate the steady state power flow in a specific hybrid HVAC/VSC-HVDC grid which is a solution that is going to be seen in the early stages of the planned supergid.

Aim: The Enhanced Liver Fibrosis (ELF) test comprises a logarithmic algorithm combining three serum markers of hepatic extracellular matrix metabolism. We aimed to evaluate the performance of ELF for the diagnosis of liver fibrosis and to compare it with that of liver stiffness measurement (LSM) by FibroScan in non‐alcoholic fatty liver disease. Methods: ELF cut‐off values for the diagnosis of advanced fibrosis were obtained using receiver operating characteristic analysis in patients with biopsy‐confirmed non‐alcoholic fatty liver disease (training set; n = 200). Diagnostic performance was analyzed in the training set and in a validation set (n = 166), and compared with that of LSM in the FibroScan cohort (n = 224). Results: The area under receiver operating characteristic curve was 0.81 for the diagnosis of advanced fibrosis, and the ELF cut‐off values were 9.34 with 90.4% sensitivity and 10.83 with 90.6% specificity in the training set, and 89.8% sensitivity and 85.5% specificity in the validation set. There was no significant difference in the area under the receiver operating characteristic curve between ELF and LSM (0.812 and 0.839). A combination of ELF (cut‐off 10.83) and LSM (cut‐off 11.45) increased the specificity to 97.9% and the positive predictive value, versus ELF alone. Sequential use of the Fibrosis‐4 index (cut‐off 2.67) and ELF (cut‐off 9.34) increased the sensitivity to 95.9%. Conclusions: ELF can identify advanced liver fibrosis in non‐alcoholic fatty liver disease, and its diagnostic accuracy is comparable to that of FibroScan. According to the clinical setting, combinations or sequential procedures using other non‐invasive tests complement the diagnostic performance of ELF for the identification of advanced fibrosis. [ABSTRACT FROM AUTHOR]

Nonparametric estimation of multivariate functions is an important problem in statistical machine learning with many applications, ranging from nonparametric regression to nonparametric graphical models. Several authors have proposed to estimate multivariate functions under the smoothing spline analysis of variance (SSANOVA) framework, which assumes that the multivariate function can be decomposed into the summation of main effects, two-way interaction effects, and higher order interaction effects. However, existing methods are not scalable to the dimension of the random variables and the order of interactions. We propose a LAyer-wiSE leaRning strategy (LASER) to estimate multivariate functions under the SSANOVA framework. The main idea is to approximate the multivariate function sequentially starting from a model with only the main effects. Conditioned on the support of the estimated main effects, we estimate the two-way interaction effects only when the corresponding main effects are estimated to be non-zero. This process is continued until no more higher order interaction effects are identified. The proposed strategy provides a data-driven approach for estimating multivariate functions under the SSANOVA framework. Our proposal yields a sequence of estimators. To establish the theoretical properties of the sequence of estimators, we establish the notion of post-selection persistency. Extensive numerical studies are performed to evaluate the performance of our algorithm. [ABSTRACT FROM AUTHOR]

The existing methods for Chinese sentiment Labeling mainly relies on the artificial sentiment corpus, but a sentiment word in the corpus may not be sentiment words in different sentences. This paper proposes a new method to label the words in the sentences by combining deep convolution neural network with sequential algorithm., We first extract the aspects comprised by words vectors, part of speech vectors, dependent syntax vectors to train the deep convolution neural network, and then employ the sequential algorithm to obtain the sentiment annotation of the sentence. Experimental results verify that our method is effective for sentiment labeling. Considering that the identification of the implicit aspects can improve the completeness of sentiment analysis, we suggest to construct the tuples including aspect, sentiment shifter, sentiment intensity, sentiment words after obtaining the sentiment labels for each word in the sentence. We develop new algorithm for implicit aspect identification by taking the two key factors of the aspects as a topic and the match degree of aspects and sentiment words, and the human language habit. The experiment demonstrates that the algorithm can effectively identify the implicit aspect. In this paper, we solve the problem of sentiment labeling and implicit aspect recognition in sentiment analysis. As a new tool for sentiment analysis, our method can be applied to the enterprise management information analysis, such as product online review, product online reputation, brand image and consumer preference management, and can also be used for the sentiment analysis of large-scale text data. [ABSTRACT FROM AUTHOR]

We consider the problem of model selection in multiple linear regression with outliers and non-normal error distributions. In this article, the robust model selection criterion is proposed based on the robust estimation method with the least absolute deviation (LAD). The proposed criterion is shown to be consistent. We suggest proposed criterion based algorithms that are suitable for a large number of predictors in the model. These algorithms select only relevant predictor variables with probability one for large sample sizes. An exhaustive simulation study shows that the criterion performs well. However, the proposed criterion is applied to a real data set to examine its applicability. The simulation results show the proficiency of algorithms in the presence of outliers, non-normal distribution, and multicollinearity. [ABSTRACT FROM AUTHOR]

Voltage source converter high voltage direct current (VSC-HVDC) transmission technology is increasing its share in AC transmission due to important advantages such as independent P and Q control and lower harmonics. Nowadays, this technology is used to connect far-away remote offshore wind-power plants to mainland networks. VSC-HVDC is the main actor in developing the future multi-terminal DC grids and supergrids. This paper proposes a sequential algorithm to calculate the steady state power flow in a specific hybrid HVAC/VSC-HVDC grid which is a solution that is going to be seen in the early stages of the planned supergid. [ABSTRACT FROM AUTHOR]