Ensemble Metamodel Research Articles

PLangRec: Deep-learning model to predict the programming language from a single line of code

Programming language detection from source code excepts remains an active research field, which has already been addressed with machine learning and natural language processing. Identifying the language of short code snippets poses both benefits and challenges across various scenarios, such as embedded code analysis, forums, Q&A systems, search engines, source code repositories, and text editors. Existing approaches for language detection typically require multiple lines or even the entire file contents. In this article, we propose a character-level deep learning model designed to predict the programming language from a single line of code. To this aim, we construct a balanced dataset comprising 434.18 million instances across 21 languages, significantly exceeding the size of existing datasets by three orders of magnitude. Leveraging this dataset, we train a deep bidirectional recurrent neural network that achieves a 95.07% accuracy and macro-F1 score for a single-line code. To predict the programming language of multiple lines (e.g., code snippets) and entire files, we build a stacking ensemble meta-model that leverages our single-line model to efficiently recognize the language of multiple lines of code. Our system outperforms the state-of-the-art approaches not only for a single line of code, but also for snippets of 5 and 10 lines and whole files of source code. We also present PLangRec, an open-source language detection system that includes our trained models. PLangRec is freely available as a user-friendly web application, a web API, and a Python desktop program.

StackedEnC-AOP: prediction of antioxidant proteins using transform evolutionary and sequential features based multi-scale vector with stacked ensemble learning

BackgroundAntioxidant proteins are involved in several biological processes and can protect DNA and cells from the damage of free radicals. These proteins regulate the body's oxidative stress and perform a significant role in many antioxidant-based drugs. The current invitro-based medications are costly, time-consuming, and unable to efficiently screen and identify the targeted motif of antioxidant proteins.MethodsIn this model, we proposed an accurate prediction method to discriminate antioxidant proteins namely StackedEnC-AOP. The training sequences are formulation encoded via incorporating a discrete wavelet transform (DWT) into the evolutionary matrix to decompose the PSSM-based images via two levels of DWT to form a Pseudo position-specific scoring matrix (PsePSSM-DWT) based embedded vector. Additionally, the Evolutionary difference formula and composite physiochemical properties methods are also employed to collect the structural and sequential descriptors. Then the combined vector of sequential features, evolutionary descriptors, and physiochemical properties is produced to cover the flaws of individual encoding schemes. To reduce the computational cost of the combined features vector, the optimal features are chosen using Minimum redundancy and maximum relevance (mRMR). The optimal feature vector is trained using a stacking-based ensemble meta-model.ResultsOur developed StackedEnC-AOP method reported a prediction accuracy of 98.40% and an AUC of 0.99 via training sequences. To evaluate model validation, the StackedEnC-AOP training model using an independent set achieved an accuracy of 96.92% and an AUC of 0.98.ConclusionOur proposed StackedEnC-AOP strategy performed significantly better than current computational models with a ~ 5% and ~ 3% improved accuracy via training and independent sets, respectively. The efficacy and consistency of our proposed StackedEnC-AOP make it a valuable tool for data scientists and can execute a key role in research academia and drug design.

Optimal redesign of coastal groundwater quality monitoring networks under uncertainty: application of the theory of belief functions.

This paper presents a new methodology for the optimal redesign of water quality monitoring networks in coastal aquifers. The GALDIT index is used to evaluate the extent and magnitude of seawater intrusion (SWI) in coastal aquifers. The weights of the GALDIT parameters are optimized using the genetic algorithm (GA). A SEAWAT-based simulation model, a spatiotemporal Kriging interpolation technique, and an artificial neural network surrogate model are then implemented to simulate total dissolved solids (TDS) concentration in coastal aquifers. To obtain more precise estimations, an ensemble meta-model is developed using the Dempster-Shafer's belief function theory (D-ST) to combine the results obtained from the three individual simulation models. The combined meta-model is then used for calculating more precise TDS concentration. Some plausible scenarios are defined for variation of water elevation and water salinity at the coastline to incorporate uncertainty through the concept of value of information (VOI). Finally, the potential wells with the highest values of information are taken into consideration to redesign coastal groundwater quality monitoring network under uncertainty. The performance of the proposed methodology is evaluated by applying it to the Qom-Kahak aquifer, north-central Iran, which is threatened by SWI. At first, the individual and ensemble simulation models are developed and validated. Then, several scenarios are defined regarding the plausible changes in TDS concentration and water level at the coastline. In the next step, the scenarios, the GALDIT-GA vulnerability map, and the VOI concept are used for redesigning the existing monitoring network. The results illustrate that the revised groundwater quality monitoring network containing 10 new sampling locations outperforms the existing one based on the VOI criterion.

Stacked KNN with hard voting predictive approach to assist hiring process in IT organizations

Effective candidates screening is critical for any IT firm as it impacts the future growth and productivity of that firm. Currently majority of these firms follow a manual approach of hiring employees which is more prone to errors and time consuming. Prime purpose of the research is to develop an intelligent predictive model to decide upon candidate’s suitability for an applied It based job. A sample size of 13,168 instances with 19 attributes of job seekers data are used for study. An ensemble predictive meta model approach is presented in the research where a stacked KNN (K-Nearest Neighbours) algorithm combined with hard voting approach is employed. The idea is to use a single unified predictive model in place of separate classification models by considering the predicted class with maximum votes against each class label. The proposed approach comprises two functional modules. ‘Stacked KNN Learner Module’ involves the use of five variants of KNN which include 1-NN, 3-NN, 5-NN, 7-NN and 9-NN. Individual prediction of the job seeker data from these variants of KNN algorithm are input to the ‘Hard Voting Ensemble Predictive unit’ which eventually aggregates majority of predicted class votes determined for each class label to generate the final output predicted class label on basis of maximum votes. The developed ensemble meta model is successfully implemented using python programming language and its performance evaluation is done using various metrics. The implemented model generated a 96.96% prediction accuracy rate. The specificity, sensitivity and f-score value recorded was 96.36%, 96.06% and 96.26% respectively. Mean Absolute Error (MAE) and Root mean Square Error (RMSE) value observed with our proposed meta model was 0.0048 and 0.0102 respectively. A comparative analysis of the proposed model was done by varying the data sample size and it gave a consistent performance throughout. It was observed that proposed ensemble model gave an impressive 96.96% mean accuracy rate and the overall performance of other variants was consistent. Performance of proposed meta model was also compared with some existing techniques like C.45, CART and K-Means. It outperformed other techniques in terms of accurately predicting the class label of candidates applying for job. This ensemble meta model can be of great help for growing IT firms and assist the organization unit in hiring right and deserving candidates in future. Proposed work can act as a decision making framework in multi sector units with enormous work force to streamline performance appraisal process thereby enabling in hiring right people for right job.

Development of a robust ensemble meta-model for prediction of salinity time series under uncertainty (case study: Talar aquifer)

The aim of this study is to develop an accurate and reliable numerical model of the coastal Talar aquifer threatened by seawater intrusion by developing an ensemble meta-model (MM). In comparison with previous methodologies, the developed model has the following superiority: (1) Its performance is enhanced by developing ensemble MMs using four different meta-modelling frameworks, i.e., artificial neural network, support vector regression, radial basis function, genetic programing and evolutionary polynomial regression; (2) The accuracy of different MMs based on 16 integration of four meta-modeling frameworks is compared; and (3) the effect of aquifer heterogeneity on the MM. The performance of the proposed MM was assessed using an illustrative case aquifer subject to seawater intrusion. The obtained results indicate that the ensemble MM that combines all four meta-modeling frameworks outperformed the GP and ANN models, with a correlation coefficient of 0.98. Moreover, the proposed MM using nonlinear-learning ensemble of SVR-EPR achieves a better and robust forecasting performance. Therefore, it can be considered as an accurate and robust simulator to predict salinity levels under different abstraction patterns in variable density flow. The result of uncertainty analyses reveals that robustness value and pumping rate are inversely proportional and scenarios with a robustness measure of about 12% are more reliable.

Ensemble of Regression-Type and Interpolation-Type Metamodels

Metamodels have become increasingly popular in the field of energy sources because of their significant advantages in reducing the computational cost of time-consuming tasks. Lacking the prior knowledge of actual physical systems, it may be difficult to find an appropriate metamodel in advance for a new task. A favorite way of overcoming this difficulty is to construct an ensemble metamodel by assembling two or more individual metamodels. Motivated by the existing works, a novel metamodeling approach for building the ensemble metamodels is proposed in this paper. By thoroughly exploring the characteristics of regression-type and interpolation-type metamodels, some useful information is extracted from the feedback of the regression-type metamodels to further improve the functional fitting capability of the ensemble metamodels. Four types of ensemble metamodels were constructed by choosing four individual metamodels. Common benchmark problems are chosen to compare the performance of the individual and ensemble metamodels. The results show that the proposed metamodeling approach reduces the risk of selecting the worst individual metamodel and improves the accuracy of the used individual metamodels.

Human resource allocation in an emergency department

Purpose The complexity and interdisciplinarity of healthcare industry problems make this industry one of the attention centers of computer-based simulation studies to provide a proper tool for interaction between decision-makers and experts. The purpose of this study is to present a metamodel-based simulation optimization in an emergency department (ED) to allocate human resources in the best way to minimize door to doctor time subject to the problem constraints which are capacity and budget. Design/methodology/approach To obtain the objective of this research, first the data are collected from a public hospital ED in Brazil, and then an agent-based simulation is designed and constructed. Afterwards, three machine-learning approaches, namely, adaptive neuro-fuzzy inference system (ANFIS), feed forward neural network (FNN) and recurrent neural network (RNN), are used to build an ensemble metamodel through adaptive boosting. Finally, the results from the metamodel are applied in a discrete imperialist competitive algorithm (ICA) for optimization. Findings Analyzing the results shows that the yellow zone section is considered as a potential bottleneck of the ED. After 100 executions of the algorithm, the results show a reduction of 24.82 per cent in the door to doctor time with a success rate of 59 per cent. Originality/value This study fulfils an identified need to optimize human resources in an ED with less computational time.

An Ensemble Meta-Modelling Approach Using the Dempster-Shafer Theory of Evidence for Developing Saltwater Intrusion Management Strategies in Coastal Aquifers

The optimum abstraction policy of coastal groundwater resources is prescribed by solving a meta-model based saltwater intrusion management model. Groundwater parameter uncertainties are explicitly incorporated into the developed meta-models in order to address the uncertainties present in coastal aquifer processes. Nevertheless, the accuracy and consequent reliability of such a management model depends upon the right choice of meta-models or a combination of meta-models. The optimal combination of meta-models, also referred to as an ensemble meta-model, can be selected by applying the Dempster-Shafer (D-S) theory of evidence. D-S evidence theory provides a platform upon which to base the selection of the best meta-model or combination of meta-models to formulate the preferred ensemble. This study demonstrates the application of D-S theory to provide an ensemble of meta-models for developing saltwater intrusion management models in coastal aquifers. The prediction accuracy of the developed ensemble meta-model is compared with that of the best standalone meta-model in the ensemble. The results confirm that the ensemble meta-model performs equally well when compared with the best meta-model in the ensemble. The developed meta-models and their ensemble are then used to develop computationally feasible multiple objective saltwater intrusion management models by utilizing an integrated simulation-optimization approach. The solution results of the management models demonstrate the superiority of the ensemble meta-model approach over standalone meta-models in obtaining Pareto optimal groundwater abstraction patterns. The evaluation of the proposed methodology is demonstrated using an illustrative multilayer coastal aquifer system subjected to groundwater parameter uncertainties.

Reliability-based NC milling parameters optimization using ensemble metamodel

In this paper, an ensemble metamodel trained by inherited latin hypercube design (ILHD) is introduced to model the effort of cutting speed and feed per tooth on the milled surface roughness and on the maximum milling force. ILHD using the initial design as inherited point is used to generate 16 combinations of milling parameters. Using these milling parameters, milling experiments are performed in Shanghai Fadal VMC4020C machining center. To further take the slight fluctuations of the milling speed and feed per tooth into consideration, the milling parameters optimization model is built by reliability-based methods. With this model, the optimal milling parameters are calculated using Monte Carlo simulation (MCS) and sequential approximation programming (SAP) algorithm. MCS is used here to conduct reliability analysis because of the nonlinear relationship between the maximum milling force and cutting parameters. Using the optimization results to perform milling verification experiments, the effectiveness of the proposed method is demonstrated, which obtains a smaller milled surface roughness with reliability satisfaction of maximum milling force.

Multi-objective robust optimization of foam-filled bionic thin-walled structures

Bio-inspired design has drawn increased attention in recent years for the excellent structural properties of biological system. In our recent work, a bionic thin-walled structure (BTS), which was inspired from the structural characteristic of horsetail, was found to have excellent crashworthiness (Yin et al., 2015) [1]. In order to further improve the crashworthiness of the BTS, a foam-filled bionic thin-walled structure (FBTS) was investigated using the software LS-DYNA in this study. And, the FBTS was optimized by a multi-objective deterministic optimization (MDO) method. The MDO result indicates that the FBTS performed better than the corresponding traditional structure. However, the deterministic optimal design is likely to become unacceptable when considering the uncertainties of design parameters. To solve this problem, a multi-objective robust optimization (MRO) method which employs ensemble metamodel, NSGA-II, “3-sigma” robust design and Monte Carlo simulation (MCS) was developed. Then, the FBTS was optimized by this MRO method. The comparison of the Pareto fronts of the MDO and MRO shows that the robust optimal FBTS is more reliable than the deterministic optimal FBTS. The robust optimal FBTS not only has excellent crashworthiness but also has high reliability. Therefore, the robust optimal FBTS is a kind of excellent and reliable energy absorber in impact engineering.

Multiobjective crashworthiness optimization design of functionally graded foam-filled tapered tube based on dynamic ensemble metamodel

Foam-filled thin-walled structures have recently gained attention with increasing interest due to their excellent energy absorption capacity. In this study, a new type of foam-filled thin-walled structure called as functionally graded foam-filled tapered tube (FGFTT) is proposed. FGFTT consists of graded density foam and thin-walled tapered tube. In order to investigate the energy absorption characteristics of FGFTTs, the numerical simulations for two kinds of FGFTTs subjected to axial dynamical loading are carried out by nonlinear finite element code LS-DYNA. In addition, a new kind of multiobjective crashworthiness optimization method employing the dynamic ensemble metamodeling method together with the multiobjective particle swarm optimization (MOPSO) algorithm is presented. This new kind of multiobjective crashworthiness optimization method is then used to implement the crashworthiness optimization design of FGFTTs. Meanwhile, the crashworthiness optimization designs of FGFTTs are implemented by using traditional multiobjective crashworthiness optimization method, which employs metamodels such as polynomial response surface (PRS), radial basis function (RBF), kriging (KRG), support vector regression (SVR) or the ensemble with the static design of experiment (DOE). Finally, by comparing the optimal designs of FGFTTs obtained by using the new multiobjective crashworthiness optimization method and the traditional one, the results show that the proposed new crashworthiness optimization method is more feasible.

Using an individual-based model with four trophic levels to model the effect of predation and competition on squid recruitment

Abstract Sinerchia, M., Field, A. J., Woods, J. D., Vallerga, S., and Hinsley, W. R. 2012. Using an individual-based model with four trophic levels to model the effect of predation and competition on squid recruitment. – ICES Journal of Marine Science, 69: 439–447. The Lagrangian Ensemble recruitment model (LERM) is the first prognostic model of fisheries recruitment based upon individuals. It incorporates five functional groups: phytoplankton (diatoms), herbivorous zooplankton (copepods), carnivorous zooplankton (squid paralarvae), and two top predators. Physiology and behaviour are described by equations derived from literature based on reproducible laboratory experiments. LERM is built using the Lagrangian Ensemble metamodel, in which the demography and biofeedback of each dynamic population are diagnostic properties, emerging from the life histories of individuals. The response of the plankton ecosystem and squid recruitment to different scenarios of exogenous forcing is investigated. Simulations were run at 41°N 27°W (Azores) under a stationary annual cycle of atmospheric forcing. The ecosystem adjusts to a stable attractor for each scenario. The emergent properties of each attractor are investigated, with focus on predation, competition for food, and spawning magnitude. Annual recruitment is a complex emergent property dependent on several factors, including food availability, predation, competition, and post-hatching growth rate, as proposed by Hjort's critical period theory, relating recruitment to predation mortality, depending on growth rate and hence food availability. The model provides a useful step towards linking small-scale processes governing the life histories of larvae and fisheries on the large scale.

The Lagrangian Ensemble metamodel for simulating plankton ecosystems

This paper presents a detailed account of the Lagrangian Ensemble (LE) metamodel for simulating plankton ecosystems. It uses agent-based modelling to describe the life histories of many thousands of individual plankters. The demography of each plankton population is computed from those life histories. So too is bio-optical and biochemical feedback to the environment. The resulting “virtual ecosystem” is a comprehensive simulation of the plankton ecosystem. It is based on phenotypic equations for individual micro-organisms. LE modelling differs significantly from population-based modelling. The latter uses prognostic equations to compute demography and biofeedback directly. LE modelling diagnoses them from the properties of individual micro-organisms, whose behaviour is computed from prognostic equations. That indirect approach permits the ecosystem to adjust gracefully to changes in exogenous forcing. The paper starts with theory: it defines the Lagrangian Ensemble metamodel and explains how LE code performs a number of computations “behind the curtain”. They include budgeting chemicals, and deriving biofeedback and demography from individuals. The next section describes the practice of LE modelling. It starts with designing a model that complies with the LE metamodel. Then it describes the scenario for exogenous properties that provide the computation with initial and boundary conditions. These procedures differ significantly from those used in population-based modelling. The next section shows how LE modelling is used in research, teaching and planning. The practice depends largely on hindcasting to overcome the limits to predictability of weather forecasting. The scientific method explains observable ecosystem phenomena in terms of finer-grained processes that cannot be observed, but which are controlled by the basic laws of physics, chemistry and biology. What-If? Prediction ( WIP), used for planning, extends hindcasting by adding events that describe natural or man-made hazards and remedial actions. Verification is based on the Ecological Turing Test, which takes account of uncertainties in the observed and simulated versions of a target ecological phenomenon. The rest of the paper is devoted to a case study designed to show what LE modelling offers the biological oceanographer. The case study is presented in two parts. The first documents the WB model (Woods & Barkmann, 1994) and scenario used to simulate the ecosystem in a mesocosm moored in deep water off the Azores. The second part illustrates the emergent properties of that virtual ecosystem. The behaviour and development of an individual plankton lineage are revealed by an audit trail of the agent used in the computation. The fields of environmental properties reveal the impact of biofeedback. The fields of demographic properties show how changes in individuals cumulatively affect the birth and death rates of their population. This case study documents the virtual ecosystem used by Woods, Perilli and Barkmann (2005; hereafter WPB); to investigate the stability of simulations created by the Lagrangian Ensemble metamodel. The Azores virtual ecosystem was created and analysed on the Virtual Ecology Workbench (VEW) which is described briefly in the Appendix.