Small Data Research Articles

Response Matching for Generating Materials and Molecules.

Diffusion models have recently emerged as powerful tools for the generation of new molecular and material structures. The key insight is that the noise in these models is related to the response of the atoms to displacement, and the denoising step is thus analogous to the geometry relaxation of atomistic systems starting from a random structure. Building on this, we present a generative method called Response Matching (RM), which leverages the fact that each stable material or molecule exists at the minimum of its potential energy surface. Any perturbation induces a response in energy and stress, driving the structure back to equilibrium. Matching this response is closely related to score matching in diffusion models. Another important aspect of state-of-the-art diffusion models is the incorporation of physical symmetries such as translation, rotation, and periodicity. RM employs a machine learning interatomic potential and random structure search as the denoising model, inherently respecting these symmetries and exploiting the locality of atomic interactions. RM handles both molecules and bulk materials under the same framework. Its efficiency and generalization are demonstrated on three systems: a small organic molecular data set, stable crystals from the Materials Project, and one-shot learning on a single diamond configuration.

A Prospective Study on Risk Prediction of Preeclampsia Using Bi-Platform Calibration and Machine Learning.

Preeclampsia is a pregnancy syndrome characterized by complex symptoms which cause maternal and fetal problems and deaths. The aim of this study is to achieve preeclampsia risk prediction and early risk prediction in Xinjiang, China, based on the placental growth factor measured using the SiMoA or Elecsys platform. A novel reliable calibration modeling method and missing data imputing method are proposed, in which different strategies are used to adapt to small samples, training data, test data, independent features, and dependent feature pairs. Multiple machine learning algorithms were applied to train models using various datasets, such as single-platform versus bi-platform data, early pregnancy versus early plus non-early pregnancy data, and real versus real plus augmented data. It was found that a combination of two types of mono-platform data could improve risk prediction performance, and non-early pregnancy data could enhance early risk prediction performance when limited early pregnancy data were available. Additionally, the inclusion of augmented data resulted in achieving a high but unstable performance. The models in this study significantly reduced the incidence of preeclampsia in the region from 7.2% to 2.0%, and the mortality rate was reduced to 0%.

Morphology, Morphogenesis and Molecular Phylogeny of Two Freshwater Ciliates (Alveolata, Ciliophora), with Description of Pseudosincirra binaria sp. nov. and Redefinition of Pseudosincirra and Perisincirra

Ciliated microeukaryotes are insufficiently investigated despite their ubiquity and ecological significance. The morphology and morphogenesis of a new Stichotrichida species, Pseudosincirra binaria sp. nov., and the known Perisincirra paucicirrata Foissner et al., 2002, are here studied using live observations and protargol staining methods. The new species is characterized by having one buccal, one parabuccal and three frontal cirri, one frontoventral row extending to the posterior half of the cell, three left and two right marginal rows and three dorsal kineties with the left kinety conspicuously bipartite, along with one caudal cirrus at the rear end of each kinety. During morphogenesis, there exist five frontal-ventral cirral anlagen with anlagen IV and V forming a frontoventral row in the proter, and four cirral anlagen with only anlage IV generating a frontoventral row in the opisthe. The anlagen for marginal rows and dorsal kineties develop intrakinetally. The new population of Perisincirra paucicirrata corresponds well with other isolates regarding morphology and cell development. Phylogenetic analyses based on small subunit ribosomal gene sequence data revealed that both Perisincirra and Pseudosincirra are deeply clustered in the clade consisting of species from the genera Deviata and Heterodeviata, supporting the placement of both genera into the family Deviatidae Foissner, 2016.

B-152 Analysis of carbonyl compounds in exhaled breath for detection of COVID-19

Abstract Background COVID-19 pandemic has caused enormous loss of lives and economic disruption. The global crisis has resulted in more than 700 million infections and 7 million deaths worldwide. Rapid screening and detection of highly infectious respiratory diseases is a key for preventing and curbing future epidemic and pandemic. The objective of this work is to develop a novel breath analysis technique for detection of COVID-19 using chemoselective capture of carbonyl compounds in exhaled breath for analysis by gas chromatography-mass spectrometry (GC-MS). Respiratory viral infection causes oxidative stress and inflammation in its host body which leads to production of higher levels of some carbonyl compounds through oxidation of lipids and could be detected in exhaled breath. This breath analysis technique can be rapidly adapted for screening infectious respiratory diseases for future epidemic and potential pandemic. Methods A total of 237 subjects were enrolled in this study. Of these, 142 were COVID-19 positive and 95 were negative. One liter exhaled breath in Tedlar bags from both COVID-19 positive and negative subjects confirmed by reverse transcriptase polymerase chain reaction between April 2022 and September 2023. The collected one liter breath samples were completely evacuated through a cartridge packed with silica particles in 2-3 minutes. Silica particles were coated with a reactive aminooxy agent for capture of carbonyl compounds in the breath samples through oximation reactions. The captured carbonyl compounds were eluted by methanol and the eluted solutions were analyzed by GC-MS for calculation of each compound concentration in the exhaled breath samples. Biostatistics was used to identify significant carbonyl compounds (defined as p value < 0.05) as markers of CPOVID-19 infection. A machine learning algorithm was implemented with the captured carbonyl compounds as input features for differentiation of COVID-19 positive from negative subjects. Results A total of more than 30 carbonyl compounds were detected in exhaled breath samples of all subjects. 10 features including 3 individual carbonyl compounds and 7 compound ratios show significant differences (p value <0.05) between COVID-19 positive and negative groups. The developed machine learning and artificial intelligence (AI) algorithms successfully separated COVOD-19 positive from negative with an accuracy of 95.4%, sensitivity of 96.5%, specificity of 93.7%, and AUROC=0.974. Conclusions The analysis of carbonyl compounds in exhaled breath by GC-MS has great potential for non-invasive detection of COVID-19. The method could be rapidly adapted for detection of other respiratory infectious diseases with a small sample size and data set for training of the machine learning and AI algorithms.

Segmentation study of nanoparticle topological structures based on synthetic data.

Nanoparticles exhibit broad applications in materials mechanics, medicine, energy and other fields. The ordered arrangement of nanoparticles is very important to fully understand their properties and functionalities. However, in materials science, the acquisition of training images requires a large number of professionals and the labor cost is extremely high, so there are usually very few training samples in the field of materials. In this study, a segmentation method of nanoparticle topological structure based on synthetic data (SD) is proposed, which aims to solve the issue of small data in the field of materials. Our findings reveal that the combination of SD generated by rendering software with merely 15% Authentic Data (AD) shows better performance in training deep learning model. The trained U-Net model shows that Miou of 0.8476, accuracy of 0.9970, Kappa of 0.8207, and Dice of 0.9103, respectively. Compared with data enhancement alone, our approach yields a 1% improvement in the Miou metric. These results show that our proposed strategy can achieve better prediction performance without increasing the cost of data acquisition.

Microstructure and properties of Cu-3Ti-0.3Cr-0.15Mg alloy designed via machine learning

With the development of information technology, machine learning (ML) has been widely applied in materials science. ML can avoid a multitude of trial and error experiments and improve the efficiency of discovering new materials. Hence, screening the alloys with excellent properties by machine learning is time-saving and effective. In this work, the major element descriptors were screened by importance analysis and correlation analysis as input features, and common ML algorithms were considered to establish ML model. The final ‘composition-process-property’ alloy design strategy was proposed and applied to guide the selection of alloy systems and composition. As a result, the Cu-3Ti-0.3Cr-0.15 Mg alloy was screened based on the extreme gradient boosting (XGBoost) model. The ultimate tensile strength and electrical conductivity of 1018 MPa and 20.1 % IACS were achieved for the designed alloy, respectively, which is well consistent with the predicted values. The ultra-high strength for the designed alloy can be mainly attributed to the precipitation strengthening. Specifically, the formation of nanoscale β′-Cu4Ti and Cr precipitates during the heat treatment plays a significant role in enhancing the strength of the alloy. This work offers new insights into the design of Cu-Ti alloys with excellent comprehensive properties by employing a machine learning model trained on small data sets, which also provides a potential to be applied to other materials systems.

On the Cauchy problem for a weakly coupled system of semi-linear σ-evolution equations with double dissipation

In this paper, we would like to consider the Cauchy problem for a multi-component weakly coupled system of semi-linear σ-evolution equations with double dissipation for any σ≥1. The first main purpose is to obtain the global (in time) existence of small data solutions in the supercritical condition by assuming additional L1 regularity for the initial data and using multi-loss of decay wisely. For the second main one, we are interested in establishing the blow-up results together with sharp estimates for lifespan of solutions in the subcritical case. The proof is based on a contradiction argument with the help of modified test functions to derive the upper bound estimates. Finally, we succeed in catching the lower bound estimate by constructing Sobolev spaces with the time-dependent weighted functions of polynomial type in their corresponding norms.

Firth-Type Penalized Methods of the Modified Poisson and Least-Squares Regression Analyses for Binary Outcomes.

The modified Poisson and least-squares regression analyses for binary outcomes have been widely used as effective multivariable analysis methods to provide risk ratio and risk difference estimates in clinical and epidemiological studies. However, there is no certain evidence that assessed their operating characteristics under small and sparse data settings and no effective methods have been proposed for these regression analyses to address this issue. In this article, we show that the modified Poisson regression provides seriously biased estimates under small and sparse data settings. In addition, the modified least-squares regression provides unbiased estimates under these settings. We further show that the ordinary robust variance estimators for both of the methods have certain biases under situations that involve small or moderate sample sizes. To address these issues, we propose the Firth-type penalized methods for the modified Poisson and least-squares regressions. The adjustment methods lead to a more accurate and stable risk ratio estimator under small and sparse data settings, although the risk difference estimator is not invariant. In addition, to improve the inferences of the effect measures, we provide an improved robust variance estimator for these regression analyses. We conducted extensive simulation studies to assess the performances of the proposed methods under real-world conditions and found that the accuracies of the point and interval estimations were markedly improved by the proposed methods. We illustrate the effectiveness of these methods by applying them to a clinical study of epilepsy.

Integrating small data and shape prior knowledge with gradient-enhanced Kriging through adaptive knowledge sampling

Integrating prior knowledge into surrogate models constitutes an advanced approach to tackling the challenges posed by small data. The existing gradient-enhanced Kriging (GEK) method can enhance accuracy by incorporating gradient points of shape prior knowledge. Nonetheless, our findings indicate that the strategy employed for setting gradient points significantly impacts accuracy, even when the number of gradient points remains constant. To enhance the efficient utilization of gradient information, we propose a novel Adaptive Knowledge Sampling (AKS) strategy with an innovative acquisition function that balances local exploitation and global exploration, effectively extracting gradient points from shape prior knowledge. The acquisition function consists of three parts: (1) The Local Information Filling Criterion (LIFC), designed to mitigate model error; (2) The Global Information Filling Criterion (GIFC), aimed at reducing model uncertainty in data-sparse areas; and (3) The Minimum Distance Constraint (MDC), implemented to prevent excessive clustering of gradient points. To strike a balance between model accuracy and computational cost, an adaptive stopping condition is introduced to regulate the number of gradient points sampled. The impact of hyperparameters in the AKS strategy was analyzed in detail, and the method’s performance was compared to existing approaches using six benchmark functions and a missile aerodynamic prediction case. The results demonstrate that AKS-GEK offers higher prediction accuracy and robustness when integrating small data with shape prior knowledge.

On the lifespan of nonzero background solutions to a class of focusing nonlinear Schrödinger equations

The global solvability in time and the potential for blow-up of solutions to non-integrable focusing nonlinear Schrödinger equations with nonzero boundary conditions at infinity present challenges that are less explored and understood compared to the case of zero boundary conditions. In this work, we address these questions by establishing estimates on the lifespan of solutions to non-integrable equations involving a general class of nonlinearities. These estimates depend on the size of the initial data, the growth of the nonlinearity, and relevant quantities associated with the amplitude of the background. The estimates provide quantified upper bounds for the minimum guaranteed lifespan of solutions. Qualitatively, for small initial data and background, these upper bounds suggest long survival times consistent with global existence of solutions. On the other hand, for larger initial data and background, the estimates indicate the potential for the intriguing phenomenon of instantaneous collapse in finite time. These qualitative theoretical results are illustrated via numerical simulations. Furthermore, importantly, the numerical findings motivate the proof of improved theoretical upper bounds that provide excellent quantitative agreement with the order of the numerically identified lifespan of solutions.

Research on multi-source information fusion tool wear monitoring based on MKW-GPR model

To enhance intelligent manufacturing capabilities and achieve the goal of online real-time high-precision monitoring of tool wear with small sample data, this paper proposes a multi-source information fusion method for turning tool wear monitoring based on the multi-kernel weighted Gaussian process regression (MKW-GPR) model. Firstly, the framework and process of multi-source information fusion for turning tool wear monitoring are established, and the experimental platform is constructed. Secondly, a full-life cutting experiment of the turning tool is conducted. During the experiment, tool wear image data and multi-sensor signal data, including cutting force and vibration signals, are collected. Then, the experimental data were processed to construct a dataset with multi-source information fusion, leading to the establishment of the MKW-GPR model. Subsequently, the experimental dataset was employed to validate the effectiveness of the turning tool wear monitoring model. The results demonstrate that, in recognizing the wear state, the MKW-GPR model exhibited superior recognition performance compared to the single kernel function Gaussian process regression model. Furthermore, in predicting the wear value, the MKW-GPR model is better than the long short-term memory (LSTM), random forest (RF), and scalable end-to-end tree boosting (XGBoost) models. During the small sample size priori point data prediction performance test, the MKW-GPR model achieved the highest accuracy, with a maximum relative error of only 2.8 % in predicting tool life during the severe wear stage. These findings substantiate the feasibility and effectiveness of the proposed turning tool wear monitoring method.

Measuring Dwarf Galaxy Intrinsic Abundance Scatter with Mid-resolution Spectroscopic Surveys: Calibrating APOGEE Abundance Errors

The first generations of stars left their chemical fingerprints on metal-poor stars in the Milky Way and its surrounding dwarf galaxies. While instantaneous and homogeneous enrichment implies that groups of conatal stars should have the same element abundances, small amplitudes of abundance scatter are seen at fixed [Fe/H]. Measurements of intrinsic abundance scatter have been made with small high-resolution spectroscopic data sets where measurement uncertainty is small compared to this scatter. In this work, we present a method to use mid-resolution survey data, which have larger errors, to make this measurement. Using APOGEE Data Release 17, we calculate the intrinsic scatter of Al, O, Mg, Si, Ti, Ni, and Mn relative to Fe for 333 metal-poor stars across six classical dwarf galaxies around the Milky Way, and 1604 stars across 19 globular clusters (GCs). We calibrate the reported abundance errors in bins of signal-to-noise ratio and [Fe/H] using a high-fidelity halo data set. Applying these calibrated errors to the APOGEE data, we find small amplitudes of average intrinsic abundance scatter in dwarf galaxies ranging from 0.03 to 0.09 dex, with a median value of 0.047 dex. For the GCs, we find intrinsic scatters ranging from 0.01 to 0.11 dex, with particularly high scatter for Al and O. Our measurements of intrinsic abundance scatter place important upper bounds, which are limited by our calibration, on the intrinsic scatter in these systems, as well as constraints on their underlying star formation history and mixing that we can look to simulations to interpret.

Research on Fracture Parameter Optimization Method Based on Generative Adversarial Network in Small Sample Size

Abstract This study addresses the optimization of fracturing parameters in the fractured gas reservoirs of the Tarim Basin, especially under the challenge of small sample sizes and high data acquisition costs. It proposes an innovative method based on Generative Adversarial Networks (GAN). Traditional fracturing technology parameter optimization often leads to model overfitting and poor generalization due to limited data. This research adopts GAN for data augmentation to generate additional training samples, thereby improving the training effect of machine learning models. The combination of genetic algorithms and model fusion techniques further optimizes production capacity prediction and fracturing parameter adjustment. Application of this method to actual data from the Tarim Basin shows that it effectively enhances the accuracy of parameter optimization and the generalization ability of the model, providing significant guidance for gas reservoir development under similar complex geological conditions.

Research on Amplification Algorithm of Small Sample Soil Composite Data Based on Probability Distribution

Abstract Analyzing and studying the data of soil compounds from 10 cities, two important features were found in this project. First, the data for all compounds in soil follow a normal distribution with different parameters. Therefore, this paper argues that small-sample soil compound amplification data should also have these two characteristics. Based on this, this paper proposes a new algorithm called PDWCC (probability distribution with correlation coefficient) for amplification of small-sample soil compound data. The algorithm combines a probability distribution with a Pearson correlation coefficient. In addition, an optimized genetic algorithm incorporating a mutation operator was introduced to improve the efficiency of data amplification. Finally, experimental comparisons were made with conventional Bootstrap and SMOTE amplification algorithms using SVM classifiers. The results show that the PDWCC amplification algorithm outperforms the other algorithms in terms of classification accuracy, recall and F1 value with mean values of 81.77%, 79.34% and 78.65%, respectively.

Evaluating the Robustness of Transfer Learning with Recipes on Small Data– Using Data of Birds as an Example

Processing small data in machine learning often leads to challenges like low accuracy and overfitting. To address these issues effectively, it is essential to assess the integrity of the underlying problem. One effective approach to tackling such challenges is to adopt a top-down strategy, focusing on adjusting and creating a suitable framework. In this paper, various techniques will be employed to fine-tune the model for optimization. Experiments will be conducted on six distinct datasets to enhance the model’s accuracy and prevent overfitting.

Small data in der linguistischen Diskursanalyse?

Quantitative approaches and big data analyses are widespread in current discourse linguistic studies and have influenced the field considerably. However, building on Fix’ ONE-text-discourse-analysis (2015) and using the example of discourses around multilingualism, this paper will try and attempt an exemplary small data analysis, using a very small corpus of just three very different data samples in order to see what kind of insights can be gained with a strictly qualitative approach. The texts that were chosen deal with discourses around multilingualism in very different ways. They include an online newspaper article, language portraits by multilingual school children and a language learning video on methods how to study a language such as German. After analysing all three artefacts and their particular value and insights when it comes to research on discursive attitudes towards multilingualism in Germany, it shall be discussed whether and how this small data approach could generally be useful in combination with other discourse linguistic methods.

Nonlinear Memory Term for Fractional Diffusion Equation

This paper examines the Cauchy problem described by the following equation: ∂tλ+1ϕ-Δϕ=∫0t(t-s)-γϕ(s,.)pds,ϕ(0,x)=ϕ0(x),ϕt(0,x)=ϕ1(x).(1)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\begin{aligned} \\partial _{t}^{\\lambda +1}\\phi -\\Delta \\phi =\\int _{0}^{t}(t-s)^{- \\gamma } \\left| \\phi (s,.) \\right| ^{p}ds,\\quad \\phi (0,x)=\\phi _{0}(x),\\quad \\phi _{t}(0,x)=\\phi _{1}(x). \\quad \\mathrm{(1)} \\end{aligned}$$\\end{document}The equation involves the Caputo fractional derivative in time, denoted as ∂tλ+1ϕ\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\partial _{t}^{\\lambda +1}\\phi$$\\end{document}. Additionally, The nonlinear term is determined by the memory term ∫0t(t-s)-γϕ(s,.)pds\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\int _{0}^{t}(t-s)^{- \\gamma } \\left| \\phi (s,.) \\right| ^{p}ds$$\\end{document}, where γ∈(0,1)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\gamma \\in (0,1)$$\\end{document}. Using the fixed point theorem, we establish the global existence of solutions to the Cauchy problem (1) for small initial data. We also investigate the impact of the nonlinearity parameter on the range of the exponent p and the estimation of the solutions.

Sound source localization via distance metric learning with regularization

Sound source localization (SSL) or simply direction of arrival (DOA) classification is an important ingredient in acoustic applications. Traditional model-based algorithms are susceptible to the effects of noise and reverberation, while data-driven deep learning algorithms maintain strong performance across a variety of acoustic circumstances, but typically require a large amount of labeled data. Nevertheless, the existing datasets for SSL are not sufficiently big and diverse to achieve the full potential of deep learning algorithms. Then, it is an imperative work to develop a non-data-hungry algorithm of SSL using small or medium data volume. To this end, we propose a regularized distance metric learning algorithm, that is, by means of the kernel method, we design a nonlinear feature transformation from two aspects: feature points and feature distributions. It transforms the data into a new feature space that brings features of the same class as close as possible and removes features of different classes as far away as possible, which can significantly improve the output of a DOA classifier that follows. Experimental results show that the proposed algorithm outperforms deep learning algorithms in diverse acoustic conditions.

Bayesian estimation and comparison of idiographic network models.

Idiographic network models are estimated on time series data of a single individual and allow researchers to investigate person-specific associations between multiple variables over time. The most common approach for fitting graphical vector autoregressive (GVAR) models uses least absolute shrinkage and selection operator (LASSO) regularization to estimate a contemporaneous and a temporal network. However, estimation of idiographic networks can be unstable in relatively small data sets typical for psychological research. This bears the risk of misinterpreting differences in estimated networks as spurious heterogeneity between individuals. As a remedy, we evaluate the performance of a Bayesian alternative for fitting GVAR models that allows for regularization of parameters while accounting for estimation uncertainty. We also develop a novel test, implemented in the tsnet package in R, which assesses whether differences between estimated networks are reliable based on matrix norms. We first compare Bayesian and LASSO approaches across a range of conditions in a simulation study. Overall, LASSO estimation performs well, while a Bayesian GVAR without edge selection may perform better when the true network is dense. In an additional simulation study, the novel test is conservative and shows good false-positive rates. Finally, we apply Bayesian estimation and testing in an empirical example using daily data on clinical symptoms for 40 individuals. We additionally provide functionality to estimate Bayesian GVAR models in Stan within tsnet. Overall, Bayesian GVAR modeling facilitates the assessment of estimation uncertainty which is important for studying interindividual differences of intraindividual dynamics. In doing so, the novel test serves as a safeguard against premature conclusions of heterogeneity. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

A Fault Diagnosis Method for Pumped Storage Unit Stator Based on Improved STFT-SVDD Hybrid Algorithm

Stator faults are one of the common issues in pumped storage generators, significantly impacting their performance and safety. To ensure the safe and stable operation of pumped storage generators, a stator fault diagnosis method based on an improved short-time Fourier transform (STFT)-support vector data description (SVDD) hybrid algorithm is proposed. This method establishes a fault model for inter-turn short circuits in the stator windings of pumped storage generators and analyzes the electrical and magnetic states associated with such faults. Based on the three-phase current signals observed during an inter-turn short circuit fault in the stator windings, the three-phase currents are first converted into two-phase currents using the principle of equal magnetic potential. Then, the STFT is applied to transform the time-domain signals of the stator’s two-phase currents into frequency-domain signals, and the resulting fault current spectrum is input into the improved SVDD network for processing. This ultimately outputs the diagnosis result for inter-turn short circuit faults in the stator windings of the pumped storage generator. Experimental results demonstrate that this method can effectively distinguish between normal and faulty states in pumped storage generators, enabling the diagnosis of inter-turn short circuit faults in stator windings with low cross-entropy loss. Through analysis, under small data sample conditions, the accuracy of the proposed method in this paper can be improved by up to 7.2%. In the presence of strong noise interference, the fault diagnosis accuracy of the proposed method remains above 90%, and compared to conventional methods, the fault diagnosis accuracy can be improved by up to 6.9%. This demonstrates that the proposed method possesses excellent noise robustness and small sample learning ability, making it effective in complex, dynamic, and noisy environments.