Robust L2 Research Articles

Predicting postoperative delirium assessed by the Nursing Screening Delirium Scale in the recovery room for non-cardiac surgeries without craniotomy: A retrospective study using a machine learning approach.

Postoperative delirium (POD) contributes to severe outcomes such as death or development of dementia. Thus, it is desirable to identify vulnerable patients in advance during the perioperative phase. Previous studies mainly investigated risk factors for delirium during hospitalization and further used a linear logistic regression (LR) approach with time-invariant data. Studies have not investigated patients' fluctuating conditions to support POD precautions. In this single-center study, we aimed to predict POD in a recovery room setting with a non-linear machine learning (ML) technique using pre-, intra-, and postoperative data. The target variable POD was defined with the Nursing Screening Delirium Scale (Nu-DESC) ≥ 1. Feature selection was conducted based on robust univariate test statistics and L1 regularization. Non-linear multi-layer perceptron (MLP) as well as tree-based models were trained and evaluated-with the receiver operating characteristics curve (AUROC), the area under precision recall curve (AUPRC), and additional metrics-against LR and published models on bootstrapped testing data. The prevalence of POD was 8.2% in a sample of 73,181 surgeries performed between 2017 and 2020. Significant univariate impact factors were the preoperative ASA status (American Society of Anesthesiologists physical status classification system), the intraoperative amount of given remifentanil, and the postoperative Aldrete score. The best model used pre-, intra-, and postoperative data. The non-linear boosted trees model achieved a mean AUROC of 0.854 and a mean AUPRC of 0.418 outperforming linear LR, well as best applied and retrained baseline models. Overall, non-linear machine learning models using data from multiple perioperative time phases were superior to traditional ones in predicting POD in the recovery room. Class imbalance was seen as a main impediment for model application in clinical practice.

Robust low tubal rank tensor recovery via [formula omitted]E criterion

Tensor analysis has received enormous attention as the increasing prevalence of high-dimensional data in science research and engineering application. Tensor recovery is an important and meaningful problem for tensor analysis. It aims to complete a tensor from an observed subset of its entries disturbed by noise. However, classical methods either develop on the second-order statistics or Lasso-type penalty, leading to them not effectively dealing with gross or dense noise effectively. To address such issues, we propose a robust tensor recovery model for simultaneously completing a low tubal rank tensor with complex noise and missing data. Based on tensor–tensor product (t-product), we first develop a tensor factor Frobenius norm to exploit the low tubal rank property which is closely related to tensor nuclear norm and tubal rank. By utilizing the robust L2 criterion, we derive the nonconvex objective function to accommodate the low tubal rank tensor. An implementable alternating minimization algorithm has been developed to estimate the low tubal rank tensor. It is worth noting that our method is able to jointly estimate the precision parameter to capture the hidden complex noise pattern. Furthermore, some convergence properties of the proposed algorithm are presented. A series of numerical experiments are conducted on both synthetic and real-world data to demonstrate the effectiveness and robustness of the proposed approach in comparison with the state-of-the-art methods.

Predicting the transmission trends of COVID-19: an interpretable machine learning approach based on daily, death, and imported cases.

COVID-19 is caused by the SARS-CoV-2 virus, which has produced variants and increasing concerns about a potential resurgence since the pandemic outbreak in 2019. Predicting infectious disease outbreaks is crucial for effective prevention and control. This study aims to predict the transmission patterns of COVID-19 using machine learning, such as support vector machine, random forest, and XGBoost, using confirmed cases, death cases, and imported cases, respectively. The study categorizes the transmission trends into the three groups: L0 (decrease), L1 (maintain), and L2 (increase). We develop the risk index function to quantify changes in the transmission trends, which is applied to the classification of machine learning. A high accuracy is achieved when estimating the transmission trends for the confirmed cases (91.5-95.5%), death cases (85.6-91.8%), and imported cases (77.7-89.4%). Notably, the confirmed cases exhibit a higher level of accuracy compared to the data on the deaths and imported cases. L2 predictions outperformed L0 and L1 in all cases. Predicting L2 is important because it can lead to new outbreaks. Thus, this robust L2 prediction is crucial for the timely implementation of control policies for the management of transmission dynamics.

Self‐concept and self‐visions in CLIL and non‐CLIL learners and their effect on motivation

AbstractTo date, most studies that explore the differences in motivation between content and language‐integrated learning (CLIL) and non‐CLIL learners have been mainly quantitative or have not controlled for differences in socio‐economic status (SES). While many researchers agree that CLIL learners tend to have greater motivation than non‐CLIL learners, there has been little explanation of the reasons behind this difference, perhaps partly because the field lacks studies from the perspective of CLIL students themselves. Based on the L2 motivational self‐system and the construct of self‐concept, this study employed a mixed‐methods approach to explore CLIL and non‐CLIL learners’ current and future self‐visions. Fifteen‐year‐old learners in the Madrid region (n = 348) completed a questionnaire that considered their SES levels as well as experiences related to the scales of English self‐concept, academic self‐concept, and teacher expectations. The same learners were then invited to take part in focus groups in which researchers delved into the reasoning behind their motivation. Despite no significant difference in SES levels between CLIL and non‐CLIL learner groups, CLIL participants are found to display a slightly more positive self‐concept and more robust ideal L2 self‐visions, which may explain the perceived greater motivation among CLIL learners to study (in) English. Furthermore, the L2 ought‐to self is found to be a secondary, yet potentially positive force toward higher motivation.

Low‐rank preserving embedding regression for robust image feature extraction

AbstractAlthough low‐rank representation (LRR)‐based subspace learning has been widely applied for feature extraction in computer vision, how to enhance the discriminability of the low‐dimensional features extracted by LRR based subspace learning methods is still a problem that needs to be further investigated. Therefore, this paper proposes a novel low‐rank preserving embedding regression (LRPER) method by integrating LRR, linear regression, and projection learning into a unified framework. In LRPER, LRR can reveal the underlying structure information to strengthen the robustness of projection learning. The robust metric L2,1‐norm is employed to measure the low‐rank reconstruction error and regression loss for moulding the noise and occlusions. An embedding regression is proposed to make full use of the prior information for improving the discriminability of the learned projection. In addition, an alternative iteration algorithm is designed to optimise the proposed model, and the computational complexity of the optimisation algorithm is briefly analysed. The convergence of the optimisation algorithm is theoretically and numerically studied. At last, extensive experiments on four types of image datasets are carried out to demonstrate the effectiveness of LRPER, and the experimental results demonstrate that LRPER performs better than some state‐of‐the‐art feature extraction methods.

μ-Synthesis-based robust L1 adaptive control for aeropropulsion system test facility

For Aeropropulsion System Test Facility (ASTF) applied in aero-engine performance testing, it is critical to accurately control the pressure and temperature of the airflow to simulate the flight environment. This article presents a μ-synthesis-based robust L1 adaptive control method for ASTF to achieve excellent control performance within a wide working envelope. A μ controller based on a reference model is first designed as the baseline, which can address the coupling effect and guarantee the desired dynamic despite uncertainties. Thereafter, an L1 adaptive compensation term is incorporated to cancel out the large (matched and unmatched) uncertainties, and then recover the baseline performance. This control scheme augments the μ controller to control the ASTF over a wide working envelope and avoids complex gain scheduling design. Furthermore, μ controller can still ensure the robust performance of the closed-loop system even if the adaptive compensation is inadequate. Numerical simulations are performed on high-fidelity nonlinear model to verify the effectiveness of the proposed method. Compared with other state-of-the-art techniques, the robust L1 adaptive control possesses better performance and robustness.

Nonverbal Intelligence Does Matter for the Perception of Second Language Sounds

Although there has been considerable research on the interplay between intelligence and second language (L2) learning, research focusing on the intelligence and L2 speech perception link is limited. The present study aims to fill this gap. The native language of the participants was Cypriot Greek and they spoke English as an L2. The participants completed a forced-choice psychoacoustic test in which they discriminated L2 sound contrasts and a nonverbal intelligence test which measured their nonverbal intelligence capacities. They were divided into two groups according to their performance in the intelligence test, namely, a low IQ and a high IQ group. The results showed that the high IQ group discriminated the majority of the L2 contrasts better than the low IQ group. In addition, the degree of perceived difficulty for most L2 contrasts differed between the two groups. It is concluded that nonverbal intelligence is associated with the discrimination of L2 sounds. This can be explained by the possibility that either intelligence triggers the more efficient functioning of other domains, such as information processing and attention, leading to increased speech perception skills, or that it directly affects the categorization of speech sounds resulting in the development of more robust L2 categories.

Dilute long period stacking/order (LPSO)-variant phases along the composition gradient in a Mg-Ho-Cu alloy

We have systematically investigated the microstructures of as-cast Mg 97.49 Ho 1.99 Cu 0.43 Zr 0.09 alloy by atomic resolution high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM), revealing the coexistence of 18R, 14H and 24R long period stacking/order (LPSO) phases with fully coherent interfaces along step-like composition gradient in a blocky intermetallic compound distributed at grain boundary. The short-range order (SRO) L1 2 -type Cu 6 Ho 8 clusters embedded across AB'C’A-stacking fault layers are directly revealed at atomic scale. Importantly, the order degree of SRO clusters in the present dilute alloy is significant lower than previous 6M and 7M in-plane order reported in ternary Mg-TM (transition metal)-RE (rare earth) alloys, which can be well matched by 9M in-plane order. This directly demonstrates that SRO in-plane L1 2 -type clusters can be expanded into more dilute composition regions bounded along the definite TM/RE ratio of 3/4. In addition, the estimated chemical compositions of solute enriched stacking fault (SESF) in all LPSO variants are almost identical with the ideal SESF composition of 9M in-plane order, regardless of the type of LPSO phases. The results further support the viewpoint that robust L1 2 -type TM 6 RE 8 clusters play an important role in governing LPSO phase formation.

Robust [formula omitted] fuzzy adaptive controller for three-phase grid-connected photovoltaic system

A robust L1 fuzzy adaptive controller is proposed in this paper for a grid-connected photovoltaic (PV) system. The latter contains a solar PV array connected to a three phase grid through a DC/AC inverter. The designed controller consists of a control law, an adaptive mechanism and a predictor. A fuzzy system is incorporated in the controller structure to achieve a good approximation of unknown nonlinear dynamics and uncertainties. Unlike conventional adaptive control strategies, the proposed control method permits fast adaptation without compromising robustness. Therefore, it can significantly improve the control performance of this type of energy conversion systems characterized by fast dynamics. Furthermore, the presented control scheme ensures the stability of the closed-loop system with bounded transient and tracking performance.

Forecasting of Steam Coal Price Based on Robust Regularized Kernel Regression and Empirical Mode Decomposition

Aiming at the problem of difficulties in modeling the nonlinear relation in the steam coal dataset, this article proposes a forecasting method for the price of steam coal based on robust regularized kernel regression and empirical mode decomposition. By selecting the polynomial kernel function, the robust loss function and L2 regular term to construct a robust regularized kernel regression model are used. The polynomial kernel function does not depend on the kernel parameters and can mine the global rules in the dataset so that improves the forecasting stability of the kernel model. This method maps the features to the high-dimensional space by using the polynomial kernel function to transform the nonlinear law in the original feature space into linear law in the high-dimensional space and helps learn the linear law in the high-dimensional feature space by using the linear model. The Huber loss function is selected to reduce the influence of abnormal noise in the dataset on the model performance, and the L2 regular term is used to reduce the risk of model overfitting. We use the combined model based on empirical mode decomposition (EMD) and auto regressive integrated moving average (ARIMA) model to compensate for the error of robust regularized kernel regression model, thus making up for the limitations of the single forecasting model. Finally, we use the steam coal dataset to verify the proposed model and such model has an optimal evaluation index value compared to other contrast models after the model performance is evaluated as per the evaluation index such as RMSE, MAE, and mean absolute percentage error.

Multichannel Alignment of S Waves

ABSTRACT High-resolution earthquake locations and structural inversions using body waves rely on precise delay-time measurements. Subsample accuracy can be realized for P waves using multichannel cross correlation (MCCC), as developed by VanDecar and Crosson (1990), which exploits redundancy in pairwise cross correlations to determine delays between similar waveforms in studies of mantle structure using teleseismic sources (common source and multiple stations) and regional studies of structure and seismicity (multiple sources and common station). For regional S waves, alignment is complicated by the additional degree of freedom in waveform polarity that is expressed for sources with different moment tensors. Here, we recast MCCC within a principal component framework and demonstrate the equivalence between maximizing waveform correlation and minimization of various singular value–based objective functions for P waves. The singular-value framework is more general and leads naturally to an MCCC linear system for S waves that possesses an order of magnitude greater redundancy than that for P waves. Robust L1 solution of the system provides an effective means of mitigating outliers at the expense of subsample precision. Residual time shifts associated with higher-order singular vectors are employed in an iterative adaptive alignment that achieves subsample resolution. We demonstrate application of the approach on a seismicity cluster within the northern Cascadia crustal fore-arc.

Multiple Taxicab Correspondence Analysis of a Survey Related to Health Services

We present an analysis of a health survey data by multiple cor respondence analysis (MCA) and multiple taxicab correspondence analysis (MTCA), MTCA being a robust L1 variant of MCA. The survey has one passive item, gender, and 22 active substantive items representing health services offered by municipal authorities; each active item has four answer categories: this service is used, never tried, tried with no access, non re sponse. We show that the first principal MTCA factor is perfectly charac terized by the sum score of the category this service is used over all service items. Further, we prove that such a sum score characterization always exists for any survey data.

On the Locus of L2 Lexical Fuzziness: Insights From L1 Spoken Word Recognition and Novel Word Learning.

The examination of how words are learned can offer valuable insights into the nature of lexical representations. For example, a common assessment of novel word learning is based on its ability to interfere with other words; given that words are known to compete with each other (Luce and Pisoni, 1998; Dahan et al., 2001), we can use the capacity of a novel word to interfere with the activation of other lexical representations as a measure of the degree to which it is integrated into the mental lexicon (Leach and Samuel, 2007). This measure allows us to assess novel word learning in L1 or L2, but also the degree to which representations from the two lexica interact with each other (Marian and Spivey, 2003). Despite the somewhat independent lines of research on L1 and L2 word learning, common patterns emerge across the two literatures (Lindsay and Gaskell, 2010; Palma and Titone, 2020). In both cases, lexicalization appears to follow a similar trajectory. In L1, newly encoded words often fail at first to engage in competition with known words, but they do so later, after they have been better integrated into the mental lexicon (Gaskell and Dumay, 2003; Dumay and Gaskell, 2012; Bakker et al., 2014). Similarly, L2 words generally have a facilitatory effect, which can, however, become inhibitory in the case of more robust (high-frequency) lexical representations. Despite the similar pattern, L1 lexicalization is described in terms of inter-lexical connections (Leach and Samuel, 2007), leading to more automatic processing (McMurray et al., 2016); whereas in L2 word learning, lack of lexical inhibition is attributed to less robust (i.e., fuzzy) L2 lexical representations. Here, I point to these similarities and I use them to argue that a common mechanism may underlie similar patterns across the two literatures.

Composite Anti-Disturbance Reference Model L2-$L_{\infty }$ Control for Helicopter Slung Load System

This work considers the anti-disturbance model reference L2- $L_{\infty }$ control for the helicopter slung load system subject to unknown disturbances. Initially, a linear model is derived based on the nonlinear dynamics of a helicopter with slung load. The uncertainty and disturbances are jointly considered in the controller design. Then, by designing a disturbance observer and using robust L2- $L_{\infty }$ control method, the state feedback control and dynamic output feedback control are respectively analyzed, and two sufficient conditions on deriving the observer gain and controller one are presented via the linear matrix inequality (LMI) form, which can guarantee the helicopter slung load system to asymptotically track the reference model with L2- $L_{\infty }$ performance. Finally, some simulation results are presented to show the effectiveness of the proposed methods.

Effects of Phonetic Training on the Discrimination of Second Language Sounds by Learners with Naturalistic Access to the Second Language.

The present study investigates the effect of high variability phonetic training (HVPT) on the discrimination of second language (L2)vowel contrasts by adult speakers who live in a countrywhere theL2 is dominant. The same subjects who participated in a previous discrimination task were trained in the discrimination of four L2vowel contrasts which wererelatively difficult for this population of learners. Both the post-test and the generalization test showed significant improvement in the discrimination of most vowel contrasts (both stressed and unstressed). The findings suggest that HVPT may facilitate the formation of robust L2 phonological representations even for learners who live and are educated in an L2-dominant environment, dissolving in that way the perceptual confusions which emerge from first language interference. Finally, important implications are made for the implementation of HVPT in L2 classrooms.

Robust stability analysis of incommensurate fractional-order systems with time-varying interval uncertainties

This paper focuses on the stability analysis of the incommensurate fractional-order systems describing by the pseudo-state-space form in the presence of interval uncertainties. By using the generalized small-gain theorem and the properties of M-matrices, a condition is achieved for robust finite-gain L2 stability of these systems. In addition, satisfying this condition ensures the asymptotic stability of uncertain fractional-order systems, based on the Caputo definition. Unlike the methods which are based on the eigenvalue argument checking, the proposed robust stability analysis method can be used for systems with irrational fractional orders and time-varying uncertainties. Compared with the Lyapunov based methods, the conservatism due to the lack of relation between stability condition and fractional-order of the system does not occur in the proposed method. An illustrative example is presented to confirm this method does not lead to conservatism generated by reformulation of the interval uncertainty, which is the basis of the commonly used methods in the literature. Finally, the suggested method is employed for robust stability checking of the Sallen–Key filter including fractional capacitors with irrational orders.

ON THE SCOPE OF OUTPUT IN SLA

AbstractFollowing calls for more modality-sensitive perspectives of SLA, this study investigated the extent to which (a) producing the second language (L2) in the oral modality impacts learner-generated noticing and L2 development of grammatical structures embedded in subsequent auditory input, and whether (b) engaging in L2 production and input processing in the written modality differentially contributes to learner noticing and L2 outcomes compared to the oral modality. Participants were beginner-level L2 Spanish learners assigned to one of three pedagogic task conditions (No-output, Speaking, Writing). Two target structures differing in their relative intrinsic salience were considered in the study. Learners’ noticing behaviors were gauged using stimulated recall protocols, and L2 grammar development was measured using pre-, post-, and delayed posttests of production and written and aural acceptability judgment. Results revealed that engaging in oral output promoted greater noticing and deeper analysis of auditory input as well as more robust L2 grammar development compared to no output. However, sustained linguistic gains on the lower-salience target structure were only observed among participants who engaged in output and input processing in the written modality.

Clinically and temporally specific diagnostic thresholds for plasma ACTH in the horse.

Pituitary pars intermedia dysfunction (PPID) is commonly investigated using plasma ACTH concentrations but problems exist with currently available diagnostic thresholds. To derive temporally specific diagnostic thresholds for equine plasma ACTH concentration to be used alongside clinical judgement in each individual week of the year and appropriate for the degree of clinical suspicion in any given case. Furthermore, to apply these thresholds to compare the prevalence of high and low ACTH in two subgroups of animals with high and low clinical suspicion of PPID. A retrospective population study examining a large laboratory database of equine plasma ACTH concentrations using an indirect approach to calculate diagnostic thresholds. Logs of plasma ACTH concentrations from 75892 individual horses were examined using robust L2 estimation of mixtures of two normal distributions in categories of each week and month of the year. Thresholds dividing the two populations of high-ACTH and low-ACTH horses were then established at different levels of sensitivity and specificity and compared with clinical subgroups of horses divided based on reported clinical signs, as having high (n=4036) or low (n=3022) clinical suspicion of PPID. For most of the year there were small interweek differences in diagnostic thresholds. However, from mid-June to early-December diagnostic thresholds showed greater interweek variability, reaching a maximum in late September and early October. Grouping of high- and low-ACTH compared favourably with grouping based on clinical signs. Given the multiple sources of diagnostic samples, pre-analytical data could not be fully verified. Diagnostic thresholds for equine plasma ACTH vary through the year. It is especially important to consider the temporally specific threshold between June and December. Different clinical thresholds can be used depending on the case circumstances and whether a false-positive or false-negative diagnosis is deemed least desirable.

Mixture robust L1 probabilistic principal component regression and soft sensor application

AbstractIn this paper, the multivariate Laplace distribution (also called L1 distribution) is adopted to construct a robust probabilistic principal component regression model (MRPPCR‐L1) under multiple operating modes. In the practical industrial chemistry process, outliers exist due to incorrect recording, disturbances, and process noises and might result in modelling distortion. To address this problem, Laplace distribution, instead of the Gaussian distribution in traditional methods, is introduced to reduce the negative influence of outliers. Moreover, probabilistic principal component regression is employed for dealing with the mixture modelling problem owing to its probabilistic property to determine the operating modes. The formulation of this approach is derived with the expectation maximum algorithm and the soft sensing model is also developed for prediction. Compared to the conventional method, a numerical example and the Tennessee Eastman process are used to demonstrate the robust modelling performance of the proposed method.

Discriminative Local Sparse Representation by Robust Adaptive Dictionary Pair Learning.

In this article, we propose a structured robust adaptive dictionary pair learning (RA-DPL) framework for the discriminative sparse representation (SR) learning. To achieve powerful representation ability of the available samples, the setting of RA-DPL seamlessly integrates the robust projective DPL, locality-adaptive SRs, and discriminative coding coefficients learning into a unified learning framework. Specifically, RA-DPL improves existing projective DPL in four perspectives. First, it applies a sparse l2,1 -norm-based metric to encode the reconstruction error to deliver the robust projective dictionary pairs, and the l2,1 -norm has the potential to minimize the error. Second, it imposes the robust l2,1 -norm clearly on the analysis dictionary to ensure the sparse property of the coding coefficients rather than using the costly l0/l1 -norm. As such, the robustness of the data representation and the efficiency of the learning process are jointly considered to guarantee the efficacy of our RA-DPL. Third, RA-DPL conceives a structured reconstruction weight learning paradigm to preserve the local structures of the coding coefficients within each class clearly in an adaptive manner, which encourages to produce the locality preserving representations. Fourth, it also considers improving the discriminating ability of coding coefficients and dictionary by incorporating a discriminating function, which can ensure high intraclass compactness and interclass separation in the code space. Extensive experiments show that our RA-DPL can obtain superior performance over other state of the arts.