Effect Of Data Size Research Articles

(Invited) Machine Learning and Fast Experimental Screening-Assisted Development of Organic Solar Cell

Non-fullerene, a small molecular electron acceptor, has substantially improved the power conversion efficiency of organic photovoltaics (OPVs).[1] However, the large structural freedom of π-conjugated polymers and molecules makes it difficult to be explored with limited resources. Machine learning, which is based on the rapidly growing artificial intelligence technology, is a high-throughput method to accelerate the speed of material design and process optimization; however, it suffers from limitations in terms of prediction accuracy, interpretability, data collection, and available data (particularly, experimental data). This recognition motivates the present review, which focuses on utilizing the experimental dataset for ML to efficiently aid OPV research. The author discusses the trends in ML-OPV publications, the NFA category, and the effects of data size and explanatory variables (fingerprints or Mordred descriptors) on the prediction accuracy and explainability, which broadens the scope of ML and would be useful for the development of next-generation solar cell materials.[2]Despite the advance of ML, the predictive accuracy of ML currently remains insufficient for the design of OPV semiconductors that exhibit a complex connectivity between chemical structure and PCE. In this study, we examined the impact of data selection and the introduction of artificially generated failure data on ML predictions of NFA solar cells. The authors demonstrated that an ML model empowered by artificially generated failure data (~0% PCE by insoluble polymers based on an inappropriate choice of solubilizing side alkyl chains) led to improved predictions.[3] This approach was validated through the synthesis and characterization of twelve polymers (benzothiadiazole, thienothiophene, or tetrazine coupled with benzodithiophene; benzobisthiazole coupled with dioxo-benzodithiophene). Our work offers a facile approach to mitigate the difficulties of the ML-driven development of OPV materials that is also readily applicable to other material science fields.Reference[1] Kranthiraja, A. Saeki, Adv. Funct. Mater. 31 (2021) 2011168[2] Miyake, A. Saeki, J. Phys. Chem. Lett. 12 (2021) 12391.[3] Miyake, K. Kranthiraja, F. Ishiwari, A. Saeki, Chem. Mater. 34 (2022) 6912.

Engineering of Substrate Tunnel of P450 CYP116B3 though Machine Learning

The combinatorial complexity of the protein sequence space presents a significant challenge for recombination experiments targeting beneficial positions. To overcome these difficulties, a machine learning (ML) approach was employed, which was trained on a limited literature dataset and combined with iterative generation and experimental data implementation. The PyPEF method was utilized to identify existing variants and predict recombinant variants targeting the substrate channel of P450 CYP116B3. Through molecular dynamics simulations, eight multiple-substituted improved variants were successfully validated. Specifically, the RMSF of variant A86T/T91H/M108S/A109M/T111P was decreased from 3.06 Å (wild type) to 1.07 Å. Additionally, the average RMSF of the variant A86T/T91P/M108V/A109M/T111P decreased to 1.41 Å, compared to the wild type’s 1.53 Å. Of particular significance was the prediction that the variant A86T/T91H/M108G/A109M/T111P exhibited an activity approximately 15 times higher than that of the wild type. Furthermore, during the selection of the regression model, PLS and MLP regressions were compared. The effect of data size and data relevance on the two regression approaches has been summarized. The aforementioned conclusions provide evidence for the feasibility of the strategy that combines ML with experimental approaches. This integrated strategy proves effective in exploring potential variations within the protein sequence space. Furthermore, this method facilitates a deeper understanding of the substrate channel in P450 CYP116B3.

Effect of data size on tooth numbering performance via artificial intelligence using panoramic radiographs.

This study aims to investigate the effect of number of data on model performance, for the detection of tooth numbering problem on dental panoramic radiographs, with the help of image processing and deep learning algorithms. The data set consists of 3000 anonymous dental panoramic X-rays of adult individuals. Panoramic X-rays were labeled on the basis of 32 classes in line with the FDI tooth numbering system. In order to examine the relationship between the number of data used in image processing algorithms and model performance, four different datasets which include 1000, 1500, 2000 and 2500 panoramic X-rays, were used. The training of the models was carried out with the YOLOv4 algorithm and trained models were tested on a fixed test dataset with 500 data and compared based on F1 score, mAP, sensitivity,precision and recall metrics. The performance of the model increased as the number of data used during the training of the model increased. Therefore, the last model trained with 2500 data showed the highest success among all the trained models. Dataset size is important for dental enumeration, and large samples should be considered as more reliable.

Comparison of machine learning approaches for radioisotope identification using NaI(TI) gamma-ray spectrum

This research aims at comparing the performance of different machine learning algorithms used for NaI(TI) gamma-ray detector based radioisotope identification. Six machine learning algorithms were implemented, including support vector machine (SVM), k-nearest neighbor (KNN), logistic regression (LR), naive Bayes (NB), decision tree (DT), and multilayer perceptron (MLP). The hyper-parameters of each model were elaborately optimized. The effects of data size, statistical fluctuation, and spectrum drift were considered. Results show that for smaller data size (5 types of radioisotopes and 6000 spectra), the support vector machine and the logistic regression classifier exhibit higher identification accuracy with less training/predicting time. Whereas for larger data size (14 types of radioisotopes corresponding to the standard IEC 62327–2017), the multilayer perceptron showed highest accuracy but required the longest time for model training. The naive Bayes classifier and the decision tree were prone to make mistakes when fluctuations and distortions were added to the spectra. The k-nearest neighbor classifier, though showing high accuracy for most data sets, consumed the longest time while making prediction.

Multivariate monthly water demand prediction using ensemble and gradient boosting machine learning techniques

Water management planning requires reliable and accurate water demand forecasting. Water demand prediction is affected by variables, such as climate, socio-economic, and demographic data. This paper investigates urban monthly average water demand prediction, using classical, ensemble, and gradient boosting-based machine learning models, using the available monthly water demand, climatic, economic, and demographic data. Three train-test data split schemes on water demand timeseries were considered to determine the effect of data size on water demand prediction. Sensitivity analysis was employed to reduce input feature dimensionality while maintaining model accuracy. A univariate timeseries (water demand only) produced R2 scores up to 0.91, which increased to 0.94 with the addition of calendar and climatic features. Increasing the training data size from 70% to 90% improved the RMSE and MAE scores by ensemble and gradient boosting methods, with the random forest and the AdaBoost models showing improvements of up to 69%. The sensitivity analysis revealed a successful input reduction scheme from a potential 17 input attributes to seven inputs. Gradient boosting models showed robust and faster execution time, especially with the increase in training data, which is attractive for medium-term urban water demand forecasting.

Machine Learning-Assisted Development of Organic Solar Cell Materials: Issues, Analyses, and Outlooks.

Nonfullerene, a small molecular electron acceptor, has substantially improved the power conversion efficiency of organic photovoltaics (OPVs). However, the large structural freedom of π-conjugated polymers and molecules makes it difficult to explore with limited resources. Machine learning, which is based on rapidly growing artificial intelligence technology, is a high-throughput method to accelerate the speed of material design and process optimization; however, it suffers from limitations in terms of prediction accuracy, interpretability, data collection, and available data (particularly, experimental data). This recognition motivates the present Perspective, which focuses on utilizing the experimental data set for ML to efficiently aid OPV research. This Perspective discusses the trends in ML-OPV publications, the NFA category, and the effects of data size and explanatory variables (fingerprints or Mordred descriptors) on the prediction accuracy and explainability, which broadens the scope of ML and would be useful for the development of next-generation solar cell materials.

Fault Location Detection with k-NN in High Voltage Cables and Effects of Data Size on Classification

Yüksek gerilim elektrik tesislerindeki kablolarda meydana gelen arızaların büyük bir kısmı, kısmi boşalma kaynaklı yalıtım sorunlarından oluşmaktadır. Bu tür arızalar, genellikle kablolar üzerindeki montaj kusurlarından, nadir olarak da üretim hatalarından kaynaklanmaktadır. Bu çalışmada, yüksek ve orta gerilim sistemlerinde kullanılan çapraz bağlı polietilen yer altı kabloları (XLPE) üzerindeki kusurların konumlarının belirlenmesi amacıyla, bir dizi deneysel çalışmalar gerçekleştirilmiştir. Laboratuvar ortamında, farklı mesafelerde kusurları bulunan aynı türde XLPE kablo numunelerine aynı orta gerilim seviyesinde gerilim belirli periyotlarda uygulanmış ve elde edilen kısmi boşalma sinyalleri zaman serisi olarak dijital ortama kaydedilmiştir. Kaydedilen veriler k En Yakın Komşuluklar (k-NN) yöntemiyle sınıflandırılmış ve sınıflandırma doğrulukları, veri azaltma yöntemi ile analiz edilmiştir. Veri azaltmadaki amaç, pek çok farklı ölçüm frekansında modelin doğru sınıflandırmalar yapabildiğini gözlemlemek ve daha düşük frekans bandında çalışan cihazlarla yapılacak ölçümlerin doğruluğunu test etmektir. Sonuçlar k-NN yönteminin, kısmi boşalma sinyallerinin incelenmesinde yüksek oranda başarılı sonuçlar verdiğini göstermekte ve verilerin büyük oranda azaltılmasının, istatistiksel öznitelikler kullanıldığında, sınıflandırma doğruluğunu etkilemediğini göstermektedir.

Prediction of stiffness degradation based on machine learning: Axial elastic modulus of [0m /90n ]s composite laminates

This paper, for the first time, proposed an axial elastic modulus degradation prediction method of [0m/90n]s cross-ply laminates using a machine learning (ML) model. The data set of the ML model is established based on the published experiments and a small amount of finite element analysis (FEA) results. The effect of data size on the accuracy of ML prediction is also discussed. The proposed ML model focuses on the process of translating a mechanical problem of damage into a non-linear regression problem of ML, and the mapping between the input and output data, which is hopefully considered for some complex mechanical problems of composites. Meanwhile, the ML method also provides accurate and efficient solution for the engineering practice.

Practical Considerations for Use of Causation Entropy in Sparsity Identification

Abstract The selection of model structure is an important step in system identification for nonlinear systems in cases where the model form is not known a priori. This process, sometimes called covariate selection or sparsity identification, involves the selection of terms in the dynamic model and is performed prior to parameter estimation. Previous work has shown the applicability of an information theory quantity known as causation entropy in performing sparsity identification. While prior work established the overall feasibility of using causation entropy to eliminate extraneous terms in a model, key questions remained regarding practical implementation. This paper builds on previous work to explore key practical considerations of causation entropy sparsity identification. First, the effect of data size is explored through both analysis and simulation, and general guidance is provided on how much data is necessary to produce accurate causation entropy estimates. Second, the effects of measurement noise and model discretization error are investigated, showing that both cause degradation of the causation entropy estimation accuracy but in opposite ways. These practical effects and trends are illustrated on several example nonlinear systems. Overall, results show that the causation entropy approach is a practical technique for sparsity identification particularly in light of the guidelines presented here for data size selection and handling of error sources.

Enhanced classification of crisis related tweets using deep learning models and word embeddings

Social media plays a crucial role during emergency events by preserving intelligence about the current condition, which may save lives. Twitter is one such powerful social media platform where information about the situational awareness is directly posted by victims or bystanders. The objective of the research is to enhance the classification of crisis related tweets by utilising the deep learning models. Our work focuses on evaluating the deep learning models, the vectorisation methods and the effect of data size on them. A multilayer perceptron (MLP), a convolutional neural network (CNN) and a long short term memory (LSTM) are employed along with the vectorisation methods (GloVe and Word2Vec), in different experiments. Based on the results pertaining to the metrics of classification and the learning graphs, the LSTM model is observed to work well. The need for measures, to improve the classification of a large twitter dataset is understood from the analysis.

Enhanced classification of crisis related tweets using deep learning models and word embeddings

Social media plays a crucial role during emergency events by preserving intelligence about the current condition, which may save lives. Twitter is one such powerful social media platform where information about the situational awareness is directly posted by victims or bystanders. The objective of the research is to enhance the classification of crisis related tweets by utilising the deep learning models. Our work focuses on evaluating the deep learning models, the vectorisation methods and the effect of data size on them. A multilayer perceptron (MLP), a convolutional neural network (CNN) and a long short term memory (LSTM) are employed along with the vectorisation methods (GloVe and Word2Vec), in different experiments. Based on the results pertaining to the metrics of classification and the learning graphs, the LSTM model is observed to work well. The need for measures, to improve the classification of a large twitter dataset is understood from the analysis.

Pushing the limits of solubility prediction via quality-oriented data selection.

SummaryAccurate prediction of the solubility of chemical substances in solvents remains a challenge. The sparsity of high-quality solubility data is recognized as the biggest hurdle in the development of robust data-driven methods for practical use. Nonetheless, the effects of the quality and quantity of data on aqueous solubility predictions have not yet been scrutinized. In this study, the roles of the size and the quality of data sets on the performances of the solubility prediction models are unraveled, and the concepts of actual and observed performances are introduced. In an effort to curtail the gap between actual and observed performances, a quality-oriented data selection method, which evaluates the quality of data and extracts the most accurate part of it through statistical validation, is designed. Applying this method on the largest publicly available solubility database and using a consensus machine learning approach, a top-performing solubility prediction model is achieved.

Batch similarity comparison of line traces based on multi-strategy machine learning

Conventional methods can scarcely achieve the rapid comparison of line traces. In this study, trace detection signals based on a single-point laser were collected and smoothed. The multi-scale wavelet coefficient of the trace detection signals was obtained after noise reduction by the dual-tree complex wavelet algorithm to minimize the adverse effects of data size on successive comparison calculations. Batch similarity comparison of trace features was achieved using multiple comparison strategies, including an optimized dynamic time regulation algorithm and the recognition of the changing rate gradient. Based on the results of boosting fusion multi-strategy comparison, optimized comparisons were achieved by machine learning, and a model for the rapid comparison of trace features was established. Finally, the viability of the proposed algorithm was verified by experiments.

Study on Fuel Utilization Dynamic model of a SOFC-GT Hybrid System Based on Deep Learning Technique

In order to perform operation management tasks, including state monitoring and control strategy optimization, of a solid oxide fuel cell-gas turbine (SOFC-GT) hybrid system, a data-driven dynamic model based on deep learning technique of long short term memory (LSTM) network is developed to predict the behaviours of fuel utilization. In addition, a LSTM model with unsupervised deep auto-encoder (DAE) method was developed to extract the feature from input data. The comparison performance between the common LSTM model and DAE-LSTM model was investigated. The results show that the DAE-LSTM model can enhance the prediction performance. Moreover, the effect of data size was investigated. The results demonstrate that the unsupervised DAE-LSTM model trained by large data size can further improve the prediction performance. The maximum error is only 0.00529, and average error decreases to 0.00025. In conclusions, the unsupervised DAE-LSTM model is an effective approach to predict dynamic behaviours.

A strategy to apply machine learning to small datasets in materials science

There is growing interest in applying machine learning techniques in the research of materials science. However, although it is recognized that materials datasets are typically smaller and sometimes more diverse compared to other fields, the influence of availability of materials data on training machine learning models has not yet been studied, which prevents the possibility to establish accurate predictive rules using small materials datasets. Here we analyzed the fundamental interplay between the availability of materials data and the predictive capability of machine learning models. Instead of affecting the model precision directly, the effect of data size is mediated by the degree of freedom (DoF) of model, resulting in the phenomenon of association between precision and DoF. The appearance of precision–DoF association signals the issue of underfitting and is characterized by large bias of prediction, which consequently restricts the accurate prediction in unknown domains. We proposed to incorporate the crude estimation of property in the feature space to establish ML models using small sized materials data, which increases the accuracy of prediction without the cost of higher DoF. In three case studies of predicting the band gap of binary semiconductors, lattice thermal conductivity, and elastic properties of zeolites, the integration of crude estimation effectively boosted the predictive capability of machine learning models to state-of-art levels, demonstrating the generality of the proposed strategy to construct accurate machine learning models using small materials dataset.

Reconstruction of Complex Directional Networks with Group Lasso Nonlinear Conditional Granger Causality

Reconstruction of networks underlying complex systems is one of the most crucial problems in many areas of engineering and science. In this paper, rather than identifying parameters of complex systems governed by pre-defined models or taking some polynomial and rational functions as a prior information for subsequent model selection, we put forward a general framework for nonlinear causal network reconstruction from time-series with limited observations. With obtaining multi-source datasets based on the data-fusion strategy, we propose a novel method to handle nonlinearity and directionality of complex networked systems, namely group lasso nonlinear conditional granger causality. Specially, our method can exploit different sets of radial basis functions to approximate the nonlinear interactions between each pair of nodes and integrate sparsity into grouped variables selection. The performance characteristic of our approach is firstly assessed with two types of simulated datasets from nonlinear vector autoregressive model and nonlinear dynamic models, and then verified based on the benchmark datasets from DREAM3 Challenge4. Effects of data size and noise intensity are also discussed. All of the results demonstrate that the proposed method performs better in terms of higher area under precision-recall curve.

Effect of uncertainties in wind speed and direction on the fatigue damage of long-span bridges

In order to estimate wind-induced fatigue damage, most methods use the probability density functions of wind speed and direction. Conventional approaches assume that these functions are error-free and provide a single estimated damage value. However, the estimated damage may contain significant errors because the probability density functions are generally inexact, and the parameters for the functions rely on limited measured data. We propose a Bayesian approach to determine the fatigue damage while considering the uncertainties in the probability model and the parameter. The proposed approach provides a distribution of damage values, which enables more comprehensive analysis of fatigue damage using bounds and exceedance probability. The approach is applied to the estimation of fatigue damage of long-span bridges. We also illustrate an additional feature of the Bayesian approach, which is to systematically update the damage index as more measurements become available over time. Effect of data size is investigated to quantify the error due to the limited measured data. The influence of uncertainties in other parameters and effects of long term wind speed variability is also included in this study.

Spirit: Security and Privacy in Real-Time Monitoring System

In this paper, we investigate the effect of data size on mobile device’s computing power for different security protocol. Based on the fact that the computing power efficiency of security protocol may depend on the data size, we propose a system prototype, Spirit, which can minimize its computation power by selecting the security protocol dynamically according to different data types which may be of different data size, such as blood pressure, brain wave, overall health status, etc..

An Effect of Data Size on Performance of Effort Estimation with Missing Data Techniques

An Effect of Data Size on Performance of Effort Estimation with Missing Data Techniques

Speaker distinguishing distances: a comparative study

Speaker discrimination is a vital aspect of speaker recognition applications such as speaker identification, verification, clustering, indexing and change-point detection. These tasks are usually performed using distance-based approaches to compare speaker models or features from homogeneous speaker segments in order to determine whether or not they belong to the same speaker. Several distance measures and features have been examined for all the different applications, however, no single distance or feature has been reported to perform optimally for all applications in all conditions. In this paper, a thorough analysis is made to determine the behavior of some frequently used distance measures, as well as features, in distinguishing speakers for different data lengths. Measures studied include the Mahalanobis distance, Kullback-Leibler (KL) distance, T 2 statistic, Hellinger distance, Bhattacharyya distance, Generalized Likelihood Ratio (GLR), Levenne distance, L 2 and L ∞ distances. The Mel-Scale Frequency Cepstral Coefficient (MFCC), Linear Predictive Cepstral Coefficients (LPCC), Line Spectral Pairs (LSP) and the Log Area Ratios (LAR) comprise the features investigated. The usefulness of these measures is studied for different data lengths. Generally, a larger data size for each speaker results in better speaker differentiating capability, as more information can be taken into account. However, in some applications such as segmentation of telephone data, speakers change frequently, making it impossible to obtain large speaker-consistent utterances (especially when speaker change-points are unknown). A metric is defined for determining the probability of speaker discrimination error obtainable for each distance measure using each feature set, and the effect of data size on this probability is observed. Furthermore, simple distance-based speaker identification and clustering systems are developed, and the performances of each distance and feature for various data sizes are evaluated on these systems in order to illustrate the importance of choosing the appropriate distance and feature for each application. Results show that for tasks which do not involve any limitation of data length, such as speaker identification, the Kullback Leibler distance with the MFCCs yield the highest speaker differentiation performance, which is comparable to results obtained using more complex state-of-the-art speaker identification systems. Results also indicate that the Hellinger and Bhattacharyya distances with the LSPs yield the best performance for small data sizes.