Sparse Data Conditions Research Articles

Deep learning model for precise prediction and design of low-melting point phthalonitrile monomers

Phthalonitrile is a widely applied resin due to its outstanding high-temperature resistance and versatility. However, the high melting point of its monomer limits its range of applications. Due to the scarcity of data, traditional deep learning models face challenges in accurately predicting the properties of phthalonitrile monomers. In this work, we propose a novel deep learning model based on techniques such as molecular similarity weighting, pre-training parameter averaging, and hybrid neural networks to address the issue of data sparsity. Our model exhibits exceptionally high accuracy for predicting the melting point of phthalonitrile monomers, e.g., the mean absolute error on the experimental dataset is 15.3 °C. To explore novel low-melting monomers, we develop a high-throughput molecule generation software and use our model to filter the virtual molecules, and ultimately obtain candidates that meet low-melting and synthesizability requirements. Moreover, we have successfully synthesized one candidate molecule, and the high consistency between our predictions and experimental measurements again demonstrates our model’s accuracy. Meanwhile, molecule dynamic simulations are conducted to provide explanations for the reduction in melting point. This study not only provides guidance for designing phthalonitrile monomers, but also proposes an innovative approach for exploring of new materials under conditions of sparse data.

Estimating substance use disparities across intersectional social positions using machine learning: An application of group-lasso interaction network.

An aim of quantitative intersectional research is to model the joint impact of multiple social positions on health risk behaviors. Although moderated multiple regression is frequently used to pursue intersectional research hypotheses, such parametric approaches may produce unreliable effect estimates due to data sparsity and high dimensionality. Machine learning provides viable alternatives, offering greater flexibility in evaluating many candidate interactions amid sparse data conditions, yet remains rarely employed. This study introduces group-lasso interaction network (glinternet), a novel machine learning approach involving hierarchical regularization, to assess intersectional differences in substance use prevalence. Utilizing variable selection and parameter stabilization functionality for main and interaction effects, glinternet was employed to examine two-way interactions between three primary social positions (gender, sexual orientation, and race) predicting heavy episodic drinking, cannabis use, and cigarette use prevalence. Analyses were conducted using the All of Us Research Program (N = 283,403), a national sample with high representation from populations historically underrepresented in biomedical research. Results were replicated using holdout cross-validation and compared against logistic regression estimates. Glinternet prevalence estimates were more stable across discovery and replication samples relative to logistic regression, particularly among sparsely represented groups. Prevalence estimates for cigarette and cannabis use were elevated among sexual minority and White cisgender women compared to heterosexual and non-White women, respectively. Glinternet may improve upon traditional moderated multiple regression methods for pursuing intersectional hypotheses by improving model parsimony and parameter stability, providing novel means for quantifying health disparities among intersectional social positions. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

A Bayesian network for simultaneous keyframe and landmark detection in ultrasonic cine

Accurate landmark detection in medical imaging is essential for quantifying various anatomical structures and assisting in diagnosis and treatment planning. In ultrasound cine, landmark detection is often associated with identifying keyframes, which represent the occurrence of specific events, such as measuring target dimensions at specific temporal phases. Existing methods predominantly treat landmark and keyframe detection as separate tasks without harnessing their underlying correlations. Additionally, owing to the intrinsic characteristics of ultrasound imaging, both tasks are constrained by inter-observer variability, leading to potentially higher levels of uncertainty. In this paper, we propose a Bayesian network to achieve simultaneous keyframe and landmark detection in ultrasonic cine, especially under highly sparse training data conditions. We follow a coarse-to-fine landmark detection architecture and propose an adaptive Bayesian hypergraph for coordinate refinement on the results of heatmap-based regression. In addition, we propose Order Loss for training bi-directional Gated Recurrent Unit to identify keyframes based on the relative likelihoods within the sequence. Furthermore, to exploit the underlying correlation between the two tasks, we use a shared encoder to extract features for both tasks and enhance the detection accuracy through the interaction of temporal and motion information. Experiments on two in-house datasets (multi-view transesophageal and transthoracic echocardiography) and one public dataset (transthoracic echocardiography) demonstrate that our method outperforms state-of-the-art approaches. The mean absolute errors for dimension measurements of the left atrial appendage, aortic annulus, and left ventricle are 2.40 mm, 0.83 mm, and 1.63 mm, respectively. The source code is available at github.com/warmestwind/ABHG.

Local coupled Markov chain model for simulating varied stratigraphy

The coupled Markov chain (CMC) model is widely employed to simulate geological uncertainty based on sparse borehole data, owing to its simplicity, parameter efficiency, and computational effectiveness. However, the existing CMC methods exhibit limitations in accurately predicting the orientation and significant dip angle changes of soil stratigraphy. These limitations stem from their reliance on a globally constant Walther constant, resulting in generated layer boundaries converging to a fixed angle. To address these issues, this study develops a local CMC model to enhance the accuracy of predicting layer sequences and describe the geological uncertainty of complex geological structures using limited borehole and surface data. In the proposed local CMC method, the global geological profile is decomposed into interconnected fragments, guided by reference boreholes. For each CMC fragment, the likelihood of observational scenarios is determined to identify the layer orientation and optimal Walther’s constant corresponding to the local geological profile. By combining the most likely realizations from all CMC fragments, the final geological profile and stratigraphic uncertainty are obtained. Illustrative case studies demonstrate that the proposed approach effectively addresses the limitations of conventional CMC models in accurately simulating significant changes in layer orientation and dip angle variations, given the same conditions of sparse borehole data. The employed localization strategy enables the prediction of geological profiles that encompass special layers such as soil interlayers, anticlines, and synclines, which cannot be well achieved by conventional CMC models. The local CMC model significantly improves the matching accuracy of geological model in situations with limited boreholes. Therefore, it plays a crucial role in the analysis of geological structures under data-scarce conditions.

3' RNA-seq is superior to standard RNA-seq in cases of sparse data but inferior at identifying toxicity pathways in a model organism.

Introduction: The application of RNA-sequencing has led to numerous breakthroughs related to investigating gene expression levels in complex biological systems. Among these are knowledge of how organisms, such as the vertebrate model organism zebrafish (Danio rerio), respond to toxicant exposure. Recently, the development of 3' RNA-seq has allowed for the determination of gene expression levels with a fraction of the required reads compared to standard RNA-seq. While 3' RNA-seq has many advantages, a comparison to standard RNA-seq has not been performed in the context of whole organism toxicity and sparse data. Methods and results: Here, we examined samples from zebrafish exposed to perfluorobutane sulfonamide (FBSA) with either 3' or standard RNA-seq to determine the advantages of each with regards to the identification of functionally enriched pathways. We found that 3' and standard RNA-seq showed specific advantages when focusing on annotated or unannotated regions of the genome. We also found that standard RNA-seq identified more differentially expressed genes (DEGs), but that this advantage disappeared under conditions of sparse data. We also found that standard RNA-seq had a significant advantage in identifying functionally enriched pathways via analysis of DEG lists but that this advantage was minimal when identifying pathways via gene set enrichment analysis of all genes. Conclusions: These results show that each approach has experimental conditions where they may be advantageous. Our observations can help guide others in the choice of 3' RNA-seq vs standard RNA sequencing to query gene expression levels in a range of biological systems.

Inferring spatial source of disease outbreaks using maximum entropy.

Mathematical modeling of disease outbreaks can infer the future trajectory of an epidemic, allowing for making more informed policy decisions. Another task is inferring the origin of a disease, which is relatively difficult with current mathematical models. Such frameworks, across varying levels of complexity, are typically sensitive to input data on epidemic parameters, case counts, and mortality rates, which are generally noisy and incomplete. To alleviate these limitations, we propose a maximum entropy framework that fits epidemiological models, provides calibrated infection origin probabilities, and is robust to noise due to a prior belief model. Maximum entropy is agnostic to the parameters or model structure used and allows for flexible use when faced with sparse data conditions and incomplete knowledge in the dynamical phase of disease-spread, providing for more reliable modeling at early stages of outbreaks. We evaluate the performance of our model by predicting future disease trajectories based on simulated epidemiological data in synthetic graph networks and the real mobility network of New York State. In addition, unlike existing approaches, we demonstrate that the method can be used to infer the origin of the outbreak with accurate confidence. Indeed, despite the prevalent belief on the feasibility of contact-tracing being limited to the initial stages of an outbreak, we report the possibility of reconstructing early disease dynamics, including the epidemic seed, at advanced stages.

Attenuated and normalized item-item product network for sequential recommendation.

Sequential recommendation has become a research trending that exploits user’s recent behaviors for recommendation. The user-item interactions contain a sequential dependency that we need to capture to better recommend. Item-item Product (IIP), which models item co-occurrence, has shown good potential by characterizing the pairwise item relationships. Generally, recent behaviors have a greater impact on the current than long-term historical behaviors. And the decaying rate of influence around infrequent behaviors is fast. However, IIP ignores such a phenomenon when considering item-item relevance and leads to suboptimal performance. In this paper, we propose an attenuated IIP mechanism which is position-aware and decays the influence of historical items at an exponential rate. Besides, In order to make up for scenarios where the influence is not in a monotonous decline trend, we add another normalized IIP mechanism to complement the attenuated IIP mechanism. It also strengthen the model’s ability in discriminating favorite items under the sparse data condition by enlarging the gap of matching degree between items. Experiments conducted on five real-world datasets demonstrate that our proposed model achieves better performance than a set of state-of-the-art sequential recommendation models.

RECOMMENDATION ALGORITHM USING DATA CLUSTERING

Recommender systems play a vital role in the marketing of various goods and services. Despite the intensive growth of the theory of recommendation algorithms and a large number of their implementations, many issues remain unresolved; in particular, scalability, quality of recommendations in conditions of sparse data, and cold start. A modified collaborative filtering algorithm based on data clustering with the dynamic determination of the number of clusters and initial centroids has been developed. Data clustering is performed using the k-means method and is applied to group similar users aimed at increase of the quality of the recommendation results. The number of clusters is calculated dynamically using the silhouette method, the determination of the initial centroids is not random, but relies on the number of clusters. This approach increases the performance of the recommender system and increases the accuracy of recommendations since the search for recommendations will be carried out within one cluster where all elements are already similar. Recommendation algorithms are software-implemented for the movie recommendation system. The software implementation of various methods that allow the user to receive a recommendation for a movie meeting their preferences is carried out: a modified algorithm, memory and neighborhood-based collaborative filtering methods. The results obtained for input data of 100, 500 and 2500 users under typical conditions, data sparsity and cold start were analyzed. The modified algorithm shows the best results – from 35 to 80 percent of recommendations that meet the user's expectations. The drop in the quality of recommendations for the modified algorithm is less than 10 per cent when the number of users increases from 100 to 2500, which indicates a good level of scalability of the developed solution. In the case of sparse data (40 percent of information is missing), the quality of recommendations is 60 percent. A low quality (35 percent) of recommendations was obtained in the case of a cold start – this case needs further investigation. Constructed algorithms can be used in rating recommender systems with the ability to calculate averaged scores for certain attributes. The modified recommendation algorithm is not tied to this subject area and can be integrated into other software systems.

Operation strategy for engineered natural ventilation using machine learning under sparse data conditions

AbstractMachine learning (ML) is a useful technique for improving building operations. However, if data can only be obtained from a target building, the data shortage will limit the use of general ML models. To overcome this issue, simplified targets and limited numbers of feature variables are required. In building engineering, building physics can be used to promote ML implementations. In this paper, a case study targeting the operation of engineered natural ventilation is performed based on energy simulations. The target issue is simplified to select the preferable opening pattern, full or half open, throughout the day. Two models using two or three feature variable types are compared. The ML method can effectively predict the correct operation scheme, even in the first year of use. The correct answer rate in the case of three variable types increases from 83% to 95% in the second year, although no significant improvement is observed in the case with two variable types. The results imply the strategy of creating a simplified model first and improving the model following data acquisition works for ML model implementation in building operations.

Operation of climate-adaptive building shells utilizing machine learning under sparse data conditions

Conventional machine learning (ML) techniques based on big data are difficult to integrate directly into building operations due to the “curse of dimensionality” caused by data sparsity . While the number of feature variables in buildings is considerable, in many cases, a reliable dataset is only obtained from the target building operation through unique features. This results in insufficient data for building an operable ML model. In this study, we proposed a methodology applying a robust algorithm and carefully selected feature variables based on building physics. An operation using climate-adaptive building shells is presented as a case study. Energy simulations utilizing a generic office building model equipped with electrochromic glasses were performed on 2016–2019 weather data from Tokyo and Fukuoka, Japan. The k -nearest neighbor algorithm was employed for the ML application because of its robustness regarding small datasets, and feature variables were prearranged and carefully chosen to set an adequate combination between the numbers of feature and objective variables. Without ML, the air conditioning system operation became unstable in intermediate seasons. The ML application successfully solved this problem; 95% of the undesired cooling operation was avoided. The results prove that a simple ML algorithm could become a better solution than a complex one in cases where building physics based on building engineers’ knowledge is effectively utilized. It expands the ML application in various building operations, even in cases that do not respond to the direct application of complex ML techniques that require large datasets, known as “big data.” • Machine learning is useful for improving the operation of climate adaptive building shells. • Conventional techniques based on big data are difficult to integrate into building operations. • The curse of dimensionality can easily occur owing to the unique features of individual buildings. • Machine learning with building physics-based variables improve electrochromic glass operations. • Building physics makes even a simple machine learning algorithm applicable to various tasks.

An Investigation of Takagi-Sugeno Fuzzy Modeling for Spatial Prediction with Sparsely Distributed Geospatial Data

Fuzzy set theory has shown potential for reducing uncertainty as a result of data sparsity and also provides advantages for quantifying gradational changes like those of pollutant concentrations through fuzzy clustering based approaches. The ability to lower the sampling frequency and perform laboratory analyses on fewer samples, yet still produce an adequate pollutant distribution map, would reduce the initial cost of new remediation projects. To assess the ability of fuzzy modeling to make spatial predictions using fewer sample points, its predictive ability was compared with the ordinary kriging (OK) and inverse distance weighting (IDW) methods under increasingly sparse data conditions. This research used a Takagi–Sugeno (TS) fuzzy modelling approach with fuzzy c-means (FCM) clustering to make spatial predictions of the lead concentrations in soil. The performance of the TS model was very dependent on the number of outliers in the respective validation set. For modeling under sparse data conditions, the TS fuzzy modeling approach using FCM clustering and constant width Gaussian shaped membership functions did not show any advantages over IDW and OK for the type of data tested. Therefore, it was not possible to speculate on a possible reduction in sampling frequency for delineating the extent of contamination for new remediation projects.

On the drawback of local detrending in universal kriging in conditions of heterogeneously spaced regional TEC data, low-order trends and outlier occurrences

The study intercompares three stochastic interpolation methods originating from the same geostatistical family: least-squares collocation (LSC) known from geodesy, as well as ordinary kriging (OKR) and universal kriging (UKR) known from geology and other geosciences. The objective of this work is to assess advantages and drawbacks of fundamental differences in modeling between these methods in imperfect data conditions. These differences primarily refer to the treatment of the reference field, commonly called ‘mean value’ or ‘trend’ in geostatistical language. The trend in LSC is determined globally before the interpolation, whereas OKR and UKR detrend the observations during the modeling process. The approach to detrending leads to the evident differences between LSC, OKR and UKR, especially in severe conditions such as far from the optimal data distribution. The theoretical comparisons of LSC, OKR and UKR often miss the numerical proof, while numerical prediction examples do not apply cross-validation of the estimates, which is proven to be a reliable measure of the prediction precision and a validation of empirical covariances. Our study completes the investigations with precise parametrization of all these methods by leave-one-out validation. It finds the key importance of the detrending schemes and shows the advantage of LSC prior global detrending scheme in unfavorable conditions of sparse data, data gaps and outlier occurrence. The test case is the modeling of vertical total electron content (VTEC) derived from GNSS station data. This kind of data is a challenge for precise covariance modeling due to weak signal at higher frequencies and existing outliers. The computation of daily set of VTEC maps using the three techniques reveals the weakness of UKR solutions with a local detrending type in imperfect data conditions.

Strategies of similarity propagation in web service recommender systems.

Recently, web service recommender systems have attracted much attention due to the popularity of Service-Oriented Computing and Cloud Computing. Memory-based collaborative filtering approaches which mainly rely on the similarity calculation are widely studied to realize the recommendation. In these research works, the similarity between two users is computed based on the QoS data of their commonly-invoked services and the similarity between two services is computed based on the common users who invoked them. However, most approaches ignore that the similarity calculation is not always accurate under a sparse data condition. To address this problem, we propose a similarity propagation method to accurately evaluate the similarities between users or services. Similarity propagation means that "if A and B are similar, and B and C are similar, then A and C will be similar to some extent". Firstly, the similarity graph of users or services is constructed according to the QoS data. Then, the similarity propagation paths between two nodes on the similarity graph are discovered. Finally, the similarity along each propagation path is measured and the indirect similarity between two users or services is evaluated by aggregating the similarities of different paths connecting them. Comprehensive experiments on real-world datasets demonstrate that our similarity propagation method can outstandingly improve the QoS prediction accuracy of memory-based collaborative filtering approaches.

Logistic Regression Under Sparse Data Conditions

The impact of sparse data conditions was examined among one or more predictor variables in logistic regression and assessed the effectiveness of the Firth (1993) procedure in reducing potential parameter estimation bias. Results indicated sparseness in binary predictors introduces bias that is substantial with small sample sizes, and the Firth procedure can effectively correct this bias.

Uncertainty quantification method for mechanical behavior of C/SiC composite and its experimental validation

C/SiC composites exhibit nonlinearity and dispersions in macroscopic constitutive behaviors, and both characteristics are dependent on external load. Previously, only uncertainties in elastic modulus have been considered, and this is not sufficient. In this paper, a characterization method for uncertainty in material behavior of a C/SiC composite is established by introducing parameter uncertainties into the constitutive model. Two types of uncertainties related to material constants and functions of the constitutive model are considered and quantified with experimental data. Tolerance-interval and optimized kernel-density methods are employed to tackle sparse data condition. The method is validated against two categories of experiments: coupon tests and tension experiments for hat-shaped components. Monte-Carlo simulation and the sparse polynomial chaos expansion method are employed to propagate the uncertainties into strain responses. The predicted uncertainties of the strains agree well with the experimental results obtained, indicating the effectiveness of the method. The method could be employed easily in uncertainty quantification of structural responses, and further, could facilitate model calibration, reliability design, and other applications for which material uncertainties are significant factors.

Groundwater DRASTIC vulnerability mapping by unsupervised and supervised techniques using a modelling strategy in two levels

Production of defensible modelling tools for aquifer vulnerability mapping under the conditions of sparse data remains topical using the DRASTIC framework. DRASTIC is the acronym for seven data layers using a prescribed scoring system in terms of rates to account for local variations and weights to account for the relative importance of data layers. Also, the term framework signifies the consensual nature of choosing the data layers without any theoretical or empirical basis. Artificial Intelligence (AI) or similar techniques are used to reduce inherent subjectivities through a strategy of unsupervised learning from data by formulating Multiple Frameworks (MF) and supervised learning from Multiple Models (MMs) without searching for any ‘superior’ model. Notably, a framework lacks the procedure for iterative training and testing, as usual in modelling procedures. In this strategy, models are organised in two hierarchical levels: Level 1 comprises the construction of MFs (two frameworks) and/or MMs (two models); Level 2 comprises a model or an algorithm to reuse the outputs from Level 1 through formulating four strategies, including a mixture of MFs/ MMs, in which Support Vector Machine (SVM) is chosen by the paper to run models at Level 2. The supervision process uses target values as a function of observed nitrate-N values. The strategy is applied to the aquifer in the Malekan region, where the past extensive agricultural practices have been transformed into intensive use of fertilisers without protecting water quality in the aquifers. Results show that AIMM/MF lead to significant improvements in delineating high vulnerability areas.

LRCS model verification based on the feature selective validation method

A feature selection validation (FSV) method was introduced to quantitatively evaluate the laser radar cross section (LRCS) model. To verify the performance of the LRCS model, we compared the agreement between the LRCS experimental data and the corresponding simulated data using the FSV method, the relative error and the root mean square error (RMSE) analysis, respectively. The experimental results demonstrate that the FSV method outperforms the relative error and the RMSE analysis in comprehensiveness and accuracy under the conditions of sparse experimental data and the angle offset of experimental curve.

Fuzzy-based assessment of groundwater intrinsic vulnerability of a volcanic aquifer in the Chilean Andean Valley.

A fuzzy logic approach has been proposed to face the uncertainty caused by sparse data in the assessment of the intrinsic vulnerability of a groundwater system with parametric methods in Las Trancas Valley, Andean Mountain, south-central Chile, a popular touristic place in Chile, but lacking of a centralized drinking and sewage water public systems; this situation is a potentially source of groundwater pollution. Based on DRASTIC, GOD, and EKv and the expert knowledge of the study area, the Mamdani fuzzy approach was generated and the spatial data were processed by ArcGIS. The groundwater system exhibited areas with high, medium, and low intrinsic vulnerability indices. The fuzzy approach results were compared with traditional methods results, which, in general, have shown a good spatial agreement even though significant changes were also identified in the spatial distribution of the indices. The Mamdani logic approach has shown to be a useful and practical tool to assess the intrinsic vulnerability of an aquifer under sparse data conditions.

Water Quality Risk Assessment for the Laoguanhe River of China Using a Stochastic Simulation Method

The Laoguanhe River is a major tributary of the Danjiang River, which is the source water source for the Danjiangkou Reservoir (i.e., the source reservoir for the middle route of the South-to-North Water Diversion Project in China). This study was intended to provide scientific decision support to help manage the water environment in the Laoguanhe River and maintain high water quality levels for the water diversion project. Risk assessment based on water quality simulation was undertaken in this research. The QUAL2Kw model was used to improve the simulations of the water quality of the Laoguanhe River under sparse data conditions. Latin hypercube sampling (LHS) was used to improve flow inputs for QUAL2Kw. The model was used to calculate the water quality risks under several scenarios based on a Markov stochastic process to reflect uncertainty. The simulation results indicated that the water quality of the downstream section of the Laoguanhe River could meet the basic requirements for exporting water to the Danjiangkou Reservoir.

Mapping potential freshwater services, and their representation within Protected Areas (PAs), under conditions of sparse data. Pilot implementation for Cambodia

Freshwater is arguably one of Earth’s most threatened natural resources, on which more than 7 billion people depend. Pressures on freshwater resources from infrastructure, resource development, agricultural pollution and deforestation are mounting, particularly in developing countries. To date, conservation responses such as Protected Areas (PAs) have not typically targeted freshwater ecosystems and their services, and thus little is known about the effectiveness of these efforts in protecting them. This paper proposes and pilots an innovative freshwater services metrics framework to quantify the representation of potential freshwater services in PAs under conditions of scarce data, with a pilot application for Cambodia. Our results indicate that conservation actions have more effectively represented potential freshwater regulation services than potential freshwater provisioning services, with major rivers remaining generally unprotected. Results from the framework are then used to propose a series of context and region specific management options to improve the conservation of freshwater services in Cambodia. There is an acute need for such management options, as the country’s food security depends largely on important freshwater ecosystems such as the Tonle Sap Lake and the deep water pools systems of the Mekong River. The framework proposed can be applied in other countries or large river basins to explore the degree of representation of freshwater services within PAs systems, under conditions of sparse data.