Real-world Experimental Data Research Articles

A novel HMM distance measure with state alignment

In this paper, we introduce a novel distance measure that conforms to the definition of a semi-distance, for quantifying the similarity between Hidden Markov Models (HMMs). This distance measure is not only easier to implement, but also accounts for state alignment before distance calculation, ensuring correctness and accuracy. Our proposed distance measure presents a significant advancement in HMM comparison, offering a more practical and accurate solution compared to existing measures. Numerical examples that demonstrate the utility of the proposed distance measure are given for HMMs with continuous state probability densities. In real-world data experiments, we employ HMM to represent the evolution of financial time series or music. Subsequently, leveraging the proposed distance measure, we conduct HMM-based unsupervised clustering, demonstrating promising results. Our approach proves effective in capturing the inherent difference in dynamics of financial time series, showcasing the practicality and success of the proposed distance measure.

Crystal Structure Determination from Powder Diffraction Patterns with Generative Machine Learning.

Powder X-ray diffraction (PXRD) is a cornerstone technique in materials characterization. However, complete structure determination from PXRD patterns alone remains time-consuming and is often intractable, especially for novel materials. Current machine learning (ML) approaches to PXRD analysis predict only a subset of the total information that comprises a crystal structure. We developed a pioneering generative ML model designed to solve crystal structures from real-world experimental PXRD data. In addition to strong performance on simulated diffraction patterns, we demonstrate full structure solutions over a large set of experimental diffraction patterns. Benchmarking our model, we predicted the structure for 134 experimental patterns from the RRUFF database and thousands of simulated patterns from the Materials Project on which our model achieves state-of-the-art 42 and 67% match rate, respectively. Further, we applied our model to determine the unreported structures of materials such as NaCu2P2, Ca2MnTeO6, ZrGe6Ni6, LuOF, and HoNdV2O8 from the Powder Diffraction File database. We extended this methodology to new materials created in our lab at high pressure with previously unsolved structures and found the new binary compounds Rh3Bi, RuBi2, and KBi3. We expect that our model will open avenues toward materials discovery under conditions which preclude single crystal growth and toward automated materials discovery pipelines, opening the door to new domains of chemistry.

Smoothing and approximation of grassland fire loading data for engineering structures by Capped Extended Support Vector Regression

This research proposes a machine learning-aided data smoothing and approximation scheme to investigate grassland fire loading for engineering structures. The investigations on grassland fire loading have significant impacts on wildfire-resistant design, building codes, regulations, and the like in specific regions. Real-world data collection systems inevitably introduce imperfections, such as noise, outliers, and the like. Furthermore, the collected data are normally discretized on time and location, and thus the information between them is completely absent. Achieving a smooth fit to data and reducing these imperfections are desired to facilitate subsequent analyses. Accordingly, a supervised machine learning algorithm is developed to smooth and approximate grassland fire loading data. Within the proposed algorithm, a convex optimization programming is established, which provides theoretical support to the accuracy of the algorithm. Then, through the proposed scheme, the grassland fire loading data can be approximated into an explicit regression model, which continuously completes the missing information during the observations. Moreover, attributed to the sparsity feature of the developed algorithm, the model can be simplified in its expression to further improve its applicability in engineering. Finally, real-world experimental data and numerical simulation results are separately investigated to demonstrate the effectiveness of the proposed scheme.

Reconstruction of particle distribution for tomographic particle image velocimetry based on unsupervised learning method

The development of deep learning has inspired some new methods to solve the 3D reconstruction problem for Tomographic Particle Image Velocimetry (Tomo-PIV). However, the supervised learning method requires a large number of data with ground truth as training information, which is very difficult to gather from experiments. Although synthetic datasets can be used as alternatives, they are still not exactly the same with the real-world experimental data. In this paper, an Unsupervised Reconstruction Technique based on U-net (UnRTU) is proposed to reconstruct volume particle distribution explicitly. Instead of using ground truth data, a projection function is used as an unsupervised loss function for network training to reconstruct particle distribution. The UnRTU was compared with some traditional algebraic reconstruction algorithms and supervised learning method using synthetic data under different particle density and noise level. The results indicate that UnRTU outperforms these traditional approaches in both reconstruction quality and noise robustness, and is comparable to the supervised learning methods AI-PR. For experimental tests, particles dispersed in cured epoxy resin are moved by an electric rail with a certain speed to obtain the ground truth data of particle velocity. Compared with other algorithms, the reconstructed particle distribution by UnRTU has the best reconstruction fidelity. And the accuracy of the 3D velocity field estimated by UnRTU is 12.9% higher than that from the traditional MLOS-MART algorithm. It demonstrates significant potential and advantages for UnRTU in 3D reconstruction of particle distribution. Finally, UnRTU was successfully applied to the high-speed planar cascade airflow field, demonstrating its applicability for measuring complex fluid flow fields at higher particle density.

Temporally aligned segmentation and clustering (TASC) framework for behavior time series analysis

Behavior exhibits a complex spatiotemporal structure consisting of discrete sub-behaviors, or motifs. Continuous behavior data requires segmentation and clustering to reveal these embedded motifs. The popularity of automatic behavior quantification is growing, but existing solutions are often tailored to specific needs and are not designed for the time scale and precision required in many experimental and clinical settings. Here we propose a generalized framework with an iterative approach to refine both segmentation and clustering. Temporally aligned segmentation and clustering (TASC) uses temporal linear alignment to compute distances between and align the recurring behavior motifs in a multidimensional time series, enabling precise segmentation and clustering. We introduce an alternating-step process: evaluation of temporal neighbors against current cluster centroids using linear alignment, alternating with selecting the best non-overlapping segments and their subsequent re-clustering. The framework is evaluated on semi-synthetic and real-world experimental and clinical data, demonstrating enhanced segmentation and clustering, offering a better foundation for consequent research. The framework may be used to extend existing tools in the field of behavior research and may be applied to other domains requiring high precision of time series segmentation.

Accurate extrinsic calibration for the invisible-light 1D laser rangefinder and camera

Abstract A combined sensor, comprising a camera and a one-dimensional laser rangefinder (1D LRF), has wide application across engineering sectors, notably in aerospace. This combined sensor is pivotal for earth observation and deep space exploration. To achieve precise and stable external parameters for this combined sensor, an accurate external calibration method is proposed. Initially, a technique for localized registration of laser spots is introduced to ensure precise determination of their positions, addressing the challenge of laser invisibility in a 1D LRF. Subsequently, a data evaluation criterion known as the data synthesis criterion is presented, addressing the issue of limited constraints in traditional calibration methods. This criterion evaluates relative errors encompassing 1D LRF ranging values, camera external parameters, and laser spot positions. Finally, based on the proposed criteria, a robust extrinsic calibration method is introduced that automatically filters observation data with significant errors and utilizes the growth rate of camera spatial resolution as the termination condition. The efficacy of the proposed method is confirmed through simulation experiments and real-world data experiments.

Uncovering wall-shear stress dynamics from neural-network enhanced fluid flow measurements

Accurate prediction and measurement of wall-shear stress dynamics in fluid flows is crucial in domains as diverse as transportation, public utility infrastructure, energy technology and human health. However, we still lack adequate experimental methods that simultaneously capture the temporal and the spatial behaviour of the wall-shear stress. In this contribution, we present a holistic approach that derives these dynamics from particle-image velocimetry (PIV) measurements using a deep optical flow estimator with physical knowledge. While the experimental measurements resemble state-of-the-art PIV set-ups, the established particle image processing is replaced by a deep neural network specifically tailored to extract velocity and wall-shear stress information. Since this WSSflow framework operates at the original image resolution, it provides the respective flow field information at a much higher spatial resolution compared with state-of-the-art PIV processing. The results show that this per-pixel approach is essential for an accurate wall-shear stress estimation. The validity and physical correctness of the derived flow quantities are demonstrated with synthetic and real-world experimental data of a turbulent channel flow, a wavy turbulent channel flow and an elastic blood vessel flow. Where baseline data are available for comparison, the instantaneous and time-averaged wall-shear stress predictions accurately follow the ground truth data.

LogicGep: Boolean networks inference using symbolic regression from time-series transcriptomic profiling data.

Reconstructing the topology of gene regulatory network from gene expression data has been extensively studied. With the abundance functional transcriptomic data available, it is now feasible to systematically decipher regulatory interaction dynamics in a logic form such as a Boolean network (BN) framework, which qualitatively indicates how multiple regulators aggregated to affect a common target gene. However, inferring both the network topology and gene interaction dynamics simultaneously is still a challenging problem since gene expression data are typically noisy and data discretization is prone to information loss. We propose a new method for BN inference from time-series transcriptional profiles, called LogicGep. LogicGep formulates the identification of Boolean functions as a symbolic regression problem that learns the Boolean function expression and solve it efficiently through multi-objective optimization using an improved gene expression programming algorithm. To avoid overly emphasizing dynamic characteristics at the expense of topology structure ones, as traditional methods often do, a set of promising Boolean formulas for each target gene is evolved firstly, and a feed-forward neural network trained with continuous expression data is subsequently employed to pick out the final solution. We validated the efficacy of LogicGep using multiple datasets including both synthetic and real-world experimental data. The results elucidate that LogicGep adeptly infers accurate BN models, outperforming other representative BN inference algorithms in both network topology reconstruction and the identification of Boolean functions. Moreover, the execution of LogicGep is hundreds of times faster than other methods, especially in the case of large network inference.

Suboptimal Trajectory Planning Technique in Real UAV Scenarios with Partial Knowledge of the Environment

In recent years, the issue of trajectory planning for autonomous unmanned aerial vehicles (UAVs) has received significant attention due to the rising demand for these vehicles across various applications. Despite advancements, real-time trajectory planning remains computationally demanding, particularly with the inclusion of 3D localization using computer vision or advanced sensors. Consequently, much of the existing research focuses on semi-autonomous systems, which rely on ground assistance through the use of external sensors (motion capture systems) and remote computing power. This study addresses the challenge by proposing a fully autonomous trajectory planning solution. By introducing a real-time path planning algorithm based on the minimization of the snap, the optimal trajectory is dynamically recalculated as needed. Evaluation of the algorithm’s performance is conducted in an unknown real-world scenario, utilizing both simulations and experimental data. The algorithm was implemented in MATLAB and subsequently translated to C++ for onboard execution on the drone.

Augmented flexible Krylov subspace methods with applications to Bayesian inverse problems

This paper presents two new augmented flexible (AF)-Krylov subspace methods, AF-GMRES and AF-LSQR, to compute solutions of large-scale linear discrete ill-posed problems that can be modeled as the sum of two independent random variables, exhibiting smooth and sparse stochastic characteristics respectively. Following a Bayesian modeling approach, this corresponds to adding a covariance-weighted quadratic term and a sparsity enforcing ℓ1 term in the original least-squares minimization scheme. To handle the ℓ1 regularization term, the proposed approach constructs a sequence approximating quadratic problems that are partially solved using augmented flexible Krylov–Tikhonov methods.Compared to other traditional methods used to solve this minimization problem, such as those based on iteratively reweighted norm schemes, the new algorithms build a single (augmented, flexible) approximation (Krylov) subspace that encodes information about the different regularization terms through adaptable “preconditioning”. The solution space is then expanded as soon as a new problem within the sequence is defined. This also allows for the regularization parameters to be chosen on-the-fly at each iteration. Compared to most recent work on generalized flexible Krylov methods [6], our methods offer theoretical assurance of convergence and a more stable numerical performance. The efficiency of the new methods is shown through a variety of experiments, including a synthetic image deblurring problem, a synthetic atmospheric transport problem, and fluorescence molecular tomography reconstructions using both synthetic and real-world experimental data.

Plasma surrogate modelling using Fourier neural operators

Predicting plasma evolution within a Tokamak reactor is crucial to realizing the goal of sustainable fusion. Capabilities in forecasting the spatio-temporal evolution of plasma rapidly and accurately allow us to quickly iterate over design and control strategies on current Tokamak devices and future reactors. Modelling plasma evolution using numerical solvers is often expensive, consuming many hours on supercomputers, and hence, we need alternative inexpensive surrogate models. We demonstrate accurate predictions of plasma evolution both in simulation and experimental domains using deep learning-based surrogate modelling tools, viz., Fourier neural operators (FNO). We show that FNO has a speedup of six orders of magnitude over traditional solvers in predicting the plasma dynamics simulated from magnetohydrodynamic models, while maintaining a high accuracy (Mean Squared Error in the normalised domain ≈10−5 ). Our modified version of the FNO is capable of solving multi-variable Partial Differential Equations, and can capture the dependence among the different variables in a single model. FNOs can also predict plasma evolution on real-world experimental data observed by the cameras positioned within the MAST Tokamak, i.e. cameras looking across the central solenoid and the divertor in the Tokamak. We show that FNOs are able to accurately forecast the evolution of plasma and have the potential to be deployed for real-time monitoring. We also illustrate their capability in forecasting the plasma shape, the locations of interactions of the plasma with the central solenoid and the divertor for the full (available) duration of the plasma shot within MAST. The FNO offers a viable alternative for surrogate modelling as it is quick to train and infer, and requires fewer data points, while being able to do zero-shot super-resolution and getting high-fidelity solutions.

A polynomial regression model based educational software tool to interpret the internal combustion engine characteristics

Technical education requires regular upgrades in pedagogical methodologies to keep up student’s skill on par with ever demanding job market. This paves the way for creating newer e-learning concepts for classroom to replace or supplement established teaching protocols. In line with this motive, this study deals with the development of an educational software tool to understand the traits of an internal combustion engine. The core of this software tool consists of polynomial regression equations, which in turn was arrived from statistical models using real world experimental data. A MATLAB-based GUI allows the operator to effortlessly interact with the software tool. Upon installation, the software requires the user to define input variables for it to automatically compute data and represent the output data in both visual and tabulated form. The advantage of three-dimensional surface plots for visual representation allows for understating the interactive effect of multiple input parameters on any given output parameter. Overall, average relative error for the model is less than 6%, thus exhibiting a good statistical fit.

Engineering punching shear strength of flat slabs predicted by nature-inspired metaheuristic optimized regression system

Reinforced concrete (RC) flat slabs, a popular choice in construction due to their flexibility, are susceptible to sudden and brittle punching shear failure. Existing design methods often exhibit significant bias and variability. Accurate estimation of punching shear strength in RC flat slabs is crucial for effective concrete structure design and management. This study introduces a novel computation method, the jellyfish-least square support vector machine (JS-LSSVR) hybrid model, to predict punching shear strength. By combining machine learning (LSSVR) with jellyfish swarm (JS) intelligence, this hybrid model ensures precise and reliable predictions. The model’s development utilizes a real-world experimental data set. Comparison with seven established optimizers, including artificial bee colony (ABC), differential evolution (DE), genetic algorithm (GA), and others, as well as existing machine learning (ML)-based models and design codes, validates the superiority of the JS-LSSVR hybrid model. This innovative approach significantly enhances prediction accuracy, providing valuable support for civil engineers in estimating RC flat slab punching shear strength.

A comparative study of methods for estimating model-agnostic Shapley value explanations

Shapley values originated in cooperative game theory but are extensively used today as a model-agnostic explanation framework to explain predictions made by complex machine learning models in the industry and academia. There are several algorithmic approaches for computing different versions of Shapley value explanations. Here, we consider Shapley values incorporating feature dependencies, referred to as conditional Shapley values, for predictive models fitted to tabular data. Estimating precise conditional Shapley values is difficult as they require the estimation of non-trivial conditional expectations. In this article, we develop new methods, extend earlier proposed approaches, and systematize the new refined and existing methods into different method classes for comparison and evaluation. The method classes use either Monte Carlo integration or regression to model the conditional expectations. We conduct extensive simulation studies to evaluate how precisely the different method classes estimate the conditional expectations, and thereby the conditional Shapley values, for different setups. We also apply the methods to several real-world data experiments and provide recommendations for when to use the different method classes and approaches. Roughly speaking, we recommend using parametric methods when we can specify the data distribution almost correctly, as they generally produce the most accurate Shapley value explanations. When the distribution is unknown, both generative methods and regression models with a similar form as the underlying predictive model are good and stable options. Regression-based methods are often slow to train but quickly produce the Shapley value explanations once trained. The vice versa is true for Monte Carlo-based methods, making the different methods appropriate in different practical situations.

Extraction of Mechanical Parameters via Molecular Dynamics Simulation: Application to Polyimides.

Polyimides feature a vast number of industrial applications due to their high thermal stability and insulation properties. These polymers exhibit an exceptional combination of thermal stability and mechanical toughness, which allows the semiconductor industry to use them as a mechanical stress buffer. Here, we perform all-atom molecular dynamics (MD) simulations for such materials to assess their predictive capability with respect to their mechanical properties. Specifically, we demonstrate that the OPLS-AA force field can be used to successfully describe an often-used polyimide (i.e., Kapton®) with respect to its Young's modulus and Poisson's ratio. Two different modes to extract these mechanical properties from MD simulations are presented. In particular, our continuous deformation mode simulations almost perfectly replicate the results from real-world experimental data and are in line with predictions using other MD force fields. Our thorough investigation of Kapton® also includes an analysis of the anisotropy of normal stresses, as well as the effect of simulation properties on the predicted Young's moduli. Furthermore, the polyimide pyromellitic dianhydride/2-(4-aminophenyl)-1H-benzimidazole-5-amine (PMDA-BIA) was investigated to draw a more thorough picture of the usability of the OPLS-AA force field for polyimides.

Pull-Back Cars: Vehicles for the Instruction of Differential Equations

This article develops a mathematical model for the motion of toy pull-back cars that can be introduced and revisited throughout an undergraduate differential equations course. This survey of the pull-back car modeling scenario enables students to connect course content with real-world experiences and experimental data, while leaving room for further exploration and advancement of the model.

On Koopman-based surrogate models for non-holonomic robots

Data-driven surrogate models of dynamical systems based on the extended dynamic mode decomposition are nowadays well-established and widely applied. Further, for non-holonomic systems exhibiting a multiplicative coupling between states and controls, the usage of bilinear surrogate models has proven beneficial. However, an analysis of the approximation quality and its dependence on different hyperparameters, including physics-motivated dictionary choices, with real-world experimental data is still missing. To close this gap and due to its high practical relevance and widespread usage in applications such as service robotics, we investigate Koopman-based surrogate modeling for a differential-drive mobile robot, also in hardware.

A Neural Network-Based Weighted Voting Algorithm for Multi-Target Classification in WSN.

One of the most important applications in the wireless sensor networks (WSN) is to classify mobile targets in the monitoring area. In this paper, a neural network(NN)-based weighted voting classification algorithm is proposed on the basis of the NN-based classifier and combined with the idea of voting strategy, which is implemented on the nodes of the WSN monitoring system by means of the "upper training, lower transplantation" approach. The performance of the algorithm is verified by using real-world experimental data, and the results show that the proposed method has a higher accuracy in classifying the target signal features, achieving an average classification accuracy of about 85% when utilizing a deep neural network (DNN) and deep belief network (DBN) as the base classifier. The experiment reveals that the NN-based weighted voting algorithm enhances the target classification accuracy by approximately 5% in comparison to the single NN-based classifier, but the memory and computation time required for the algorithm to run are also increased at the same time. Compared to the FFNN classifier, which exhibited the highest classification accuracy among the four selected methods, the algorithm achieves an improvement of approximately 8.8% in classification accuracy. However, it incurs greater overhead time to run.

Prediction of Concrete's Compressive Strength via Artificial Neural Network Trained on Synthetic Data

Predicting concrete compressive strength using machine learning techniques has attracted the focus of many studies in recent years. Typically, given concrete mix ingredients, a machine learning model is trained on experimental data to predict properties of hardened concrete, such as compressive strength at 28 days. This study used computer-generated mix design data that contained mixed ingredients along with the corresponding theoretical strength of each mix to train a neural network and then test them on real-world experimental data. The developed model was able to predict the compressive strength of concrete specimens at 28 days with an R-value of 0.80. Furthermore, increasing the synthetic dataset increased the performance of the model to a point beyond which it started to decrease. The proposed sustainability-promoting method emphasizes the effectiveness of using synthetic data to train machine learning models that yield insightful predictions with acceptable accuracy.

Causal relationship analysis of high-dimensional time series based on quantile factor model

Finding the causal relationship between variables from observed data is a key issue in many decision analytics researches. Because the traditional Granger causality model is affected by the curse of dimensionality, it is difficult to accurately find causality in high-dimensional time series. To address this problem, we propose a new Granger causality analysis method based on quantile factor model, named QFM-CGC. First, QFM-CGC adopts Akaike information criterion to select models, which avoids setting the lag order relying on human intervention. Then, the quantile factor model is established to reduce the dimensionality of conditional variables in a VAR (Vector Auto-Regressive) model, further reducing the number of coefficients. After that, the reduced-dimensional VAR model is used for a conditional Granger causality analysis. Finally, in order to identify the connectivity structure between the underlying system and observation time series, Monte Carlo simulation is applied to evaluate the performance of different methods. Experiments compared with classical methods on linear simulation systems in different dimensions and two realistic datasets, all demonstrate the effectiveness of QFM-CGC. In simulation experiments, the method proposed in this article has improved the accuracy of identifying causal relationships by an average of 6% and 3.46%. In two real-world data experiments, compared with the optimal comparison methods, QFM-CGC reduced the prediction error of causal relationship sequences.