Learner Model Research Articles

Utilizing machine learning models to estimate energy savings from an industrial energy system

Energy audits are an important part of reducing energy usage, costs, and carbon emissions, but there have been discrepancies in the quality of audits depending upon the auditor, which can negatively affect the impacts and credibility of the energy assessment. In this paper, historical energy auditing data from a U.S. Department of Energy sponsored research program was gathered and analyzed with a machine-learning algorithm to predict demand savings from a compressed air system assessment recommendation in industrial manufacturing facilities. Different energy auditors calculate savings for repairing leaks in compressed air systems in various ways, so the energy demand savings have been calculated differently throughout the historical assessment recommendations. Machine learning models are utilized in order to enhance the accuracy of the existing practice and reduce variations resulting from the abovementioned discrepancies. A large set of historical assessment recommendation data was used to train five unique machine learning models. Four base learner models and one metalearner model were devised and compared. Results showed that the distributed random forest model best predicted compressed air energy demand savings against the new scenarios within an error of 17%. This indicates that the distributed random forest model can more accurately quantify savings from repairing leaks in compressed air systems. In addition, the results from this study provide insight into the important factors contributing to leaks in the compressed air systems and why it is crucial to repair those leaks regularly to save money and energy while decreasing emissions.

Lifetime prediction of electronic devices based on the P-stacking machine learning model

Nowadays the data-driven artificial intelligence (AI) and machine learning (ML) provide novel approaches for the effective lifetime prediction of the electronic devices with complicated mechanisms and multiple controlling factors. However, the existing ML prediction models cannot process both high accuracy and high efficiency at the meantime. To address this problem, this paper proposes the new P-Stacking ML algorithm model, which combines the Pearson correlation analysis and the Stacking multi-model fusion. The Pearson correlation analysis is first applied to select the multiple base learner models with weak correlations for the following Stacking multi-model fusion. Two neutral network layers are built to perform the Stacking algorithm. In this work, two distinct types of electronic devices are utilized to verify the effectiveness of the P-Stacking ML model, which are the IGBT devices and the lithium-ion batteries respectively. For the IGBT lifetime prediction, the results have shown that compared with the long-term and short-term memory neural network (LSTM) model in the previous literature, the mean square error (MSE) of the P-Stacking ML model is improved by 6 %, and the average model training time is reduced by 90 %. Moreover, for the lithium batteries, the lifetime prediction accuracy of the P-Stacking model is increased by 65 %, and the training time is decreased by 90 %. The results demonstrated that the P-Stacking ML model can significantly improve both the prediction accuracy and efficiency simultaneously.

Güneş ve Rüzgar Enerjisinden Elektrik Üretiminin Gün Öncesi Piyasa Takas Fiyatlarına ve Fiyat Volatilitesine Etkisi: Türkiye Örneği

Solar and wind generation are the primary tools to prevent climate change and high carbon emissions. Due to their intermittent generation characteristics, solar and wind power plants have a different impact on the market-clearing price formation compared to conventional generation sources. The paper investigates the solar and wind generation effect on the day ahead market-clearing price in Turkey between the 2016 and 2022. We used a 2nd order polynomial learner model to analyze the impact of solar and wind generation level on the market-clearing price level and volatility. We find that Dutch TTF increases market-clearing price with a coefficient of 0.24. An increase in wind and solar generation reduce the market-clearing price. Solar generation is ineffective on the market-clearing price below a certain demand level. Wind generation reduces market-clearing price with a 37.78 coefficient at low demand levels and a 6.55 coefficient at high demand levels. Solar generation has a price-reducing effect with 5.55 at high demand levels. Finally, Dutch TTF and wind generation increased volatility with coefficients of 0.04 and 0.69; solar generation reduced volatility with a coefficient of 0.83.

Borzuya, The physician: A curriculum output of Khosravani Education

Introduction: Borzuya, the physician, a sage as well, is one of the prominent figures of Shahnameh in Sassanid era especially in Nushirvan court. In seeking of the immortality tree, he made a voyage to India. Such a burden to bear for memorization and its translation to Pahlavi and then dedication the work made him immortal owing to the allocation of the first chapter as Borzuya, the sage.
 Methods: Regarding a method fusion of history, analysis, and education, the germane sources were reflected upon and some presuppositions were critically pondered as the similarity of Borzuya and Bozorgmehr. However, these names are two distinguished figures in Shahnameh, a physician and a minister, respectively. Regarding, the concept of development is not allocated to one facet, but it is multifaceted. Also it must be noted that the true outreach won’t come but with comprehensive and true thinking, design, edit, execution and assessment.
 Results: First, the education of Ancient Persia is categorized as an educational archetype due to interaction of Iranian civilization with its east and west civilizations. Second, the educational implications of Shahnameh can be represented in different levels; basic (art and wisdom education) and supreme (intuition and immortality education).
 Conclusion: One of the principles of education including medical education is allocated to the role model presentation for the students. Owing to the personality of Bozuya who was a famous physician and furthermore he was an expert in philosophy, linguistics, literature, and logic, he can be presented as an appropriate role model for any contemporary learner.

Modeling of microbial fuel cell power generation using machine learning-based super learner algorithms

Electricity generation from microbial fuel cells (MFCs) is a potential environment-friendly technology. This study provides Bayesian Algorithm (BA) based Support Vector Regression (SVR) and Boosted Regression Tree (BRT) as prospective super learner modeling tools (BA-SVR, BA-BRT) for predictions of electricity production from MFCs. The membrane thickness, external resistance, and anode area were considered independent variables, while power generation was taken as a response variable. The key novelties of this study include (i) hybridization of BA with SVR and BRT (separately) for forecasting power generation from fuel cells for the first time, (ii) performance comparison of the developed models (BA-SVR and BA-BRT) with the existing Response Surface Methodology (RSM) based on the coefficient of determination (R2), relative error (RE), mean absolute error (MAE), mean absolute percentage error (MAPE), root mean square error (RMSE), and computing efficiency, and the (iii) analysis of the models’ robustness by utilizing Gaussian white noise. Based on the performance indicators, the proposed super leaner models showed excellent performance compared to the existing M.J. Salar-García et al. RSM model. The BA-SVR model provided the lowest errors (MAE of 2.94, RSME of 7.2926, MAPE of 13.8341) with the highest R2 of 0.9981, compared to the BA-BRT and RSM models. The proposed BA-SVR model showed superior performance to the RSM and BA-BRT models in predicting the MFCs’ power generation, with a performance improvement of more than 90% regarding MAPE, as an example. The future prediction and high robustness of the proposed super learner model would ensure quick estimation for maximization of electricity generation that may lead to reducing massive lab trials and saving resources.

A Personalized Learning Path Recommendation Method Incorporating Multi-Algorithm

In this era of intelligence, the learning methods of learners have substantially changed. Many learners choose to learn through online education platforms. Although learners may enjoy more high-quality educational resources, when they are faced with an abundance of resource information, they are prone to become lost in knowledge, among other problems. To solve this problem, a multi-algorithm collaborative, personalized, learning path recommendation model is proposed to provide learning guidance for learners of online learning platforms. First, the learner model is constructed from four perspectives: cognitive level, learning ability, learning style, and learning intensity. Second, the association rule algorithm is employed to generate a sequence of knowledge points and to plan the learning sequence of knowledge points for learners. Last, the swarm intelligence algorithm is utilized to ensure that each knowledge point is matched with personalized learning resources with a higher degree of adaptability so that learners can learn using a more targeted approach. The experimental results show that the research results of this paper can, to a certain extent, recommend ideal learning paths to target users, effectively improve the accuracy of recommended resources, and thus improve the learning quality and learning effect of users.

Evaluation of stacked ensemble model performance to predict clinical outcomes: A COVID-19 study

BackgroundThe application of machine learning (ML) to analyze clinical data with the goal to predict patient outcomes has garnered increasing attention. Ensemble learning has been used in conjunction with ML to improve predictive performance. Although stacked generalization (stacking), a type of heterogeneous ensemble of ML models, has emerged in clinical data analysis, it remains unclear how to define the best model combinations for strong predictive performance. This study develops a methodology to evaluate the performance of “base” learner models and their optimized combination using “meta” learner models in stacked ensembles to accurately assess performance in the context of clinical outcomes. MethodsDe-identified COVID-19 data was obtained from the University of Louisville Hospital, where a retrospective chart review was performed from March 2020 to November 2021. Three differently-sized subsets using features from the overall dataset were chosen to train and evaluate ensemble classification performance. The number of base learners chosen from several algorithm families coupled with a complementary meta learner was varied from a minimum of 2 to a maximum of 8. Predictive performance of these combinations was evaluated in terms of mortality and severe cardiac event outcomes using area-under-the-receiver-operating-characteristic (AUROC), F1, balanced accuracy, and kappa. ResultsThe results highlight the potential to accurately predict clinical outcomes, such as severe cardiac events with COVID-19, from routinely acquired in-hospital patient data. Meta learners Generalized Linear Model (GLM), Multi-Layer Perceptron (MLP), and Partial Least Squares (PLS) had the highest AUROC for both outcomes, while K-Nearest Neighbors (KNN) had the lowest. Performance trended lower in the training set as the number of features increased, and exhibited less variance in both training and validation across all feature subsets as the number of base learners increased. ConclusionThis study offers a methodology to robustly evaluate ensemble ML performance when analyzing clinical data.

Predicting Miscarriage and Stillbirth Using Weighted Ensemble Machine Learning [ID: 1338167

INTRODUCTION: Prediction of miscarriage and stillbirth remains a clinical challenge. Prior efforts to use machine learning tools have not used an ensemble weighted machine learning approach, called “super learner,” which offers the opportunity to improve prediction performance by aggregating the outputs of constituent machine learning models and preferentially weighting the highest-performing models. METHODS: Data were obtained from hospital-wide electronic health records from a large academic institution. The sample comprised 13,337 patients who delivered between 2008 and 2019, 6,932 of whom experienced a miscarriage or stillbirth. Phecodes for ICD-9-CM and ICD-10-CM were used to define miscarriage and stillbirth and create comorbidity categories. The constituent models of the super learner were XGBoost, random forest, a regularized generalized linear model (both lasso and ridge regressions), and a support vector machine. The objective of this study was to develop a super learner algorithm to predict miscarriage and stillbirth. RESULTS: The super learner model predicted miscarriage and stillbirth classification with an area under the receiver operating characteristic curve of 0.94 and an accuracy of 92%. It used two models: random forest, weighted at 73%, and SVM, weighted at 27%. The most highly weighted predictors were amniotic cavity abnormalities, pelvic soft tissue abnormalities, and preeclampsia. CONCLUSION: A super learner performs comparably to other models in the literature. External validation is warranted. The promising results suggest that a super learner model can be used as a clinical decision support tool, supplementing clinical judgement in predicting miscarriage and stillbirth.

A comparative study of characteristics and preferences to learner models in educational adaptive hypermedia systems

Educational adaptive hypermedia systems are online learning systems that aim to tailor the learning experience to the characteristics and needs of individual learners. These systems use a variety of techniques, including data analysis, machine learning, and dynamic adaptive user interface, to provide more effective and personalized learning for each learner. The learner models in their systems are designed to help them better understand the learner and tailor their learning experience, to do this their models take into account the characteristics of the learner, such as their learning style, education level, prior knowledge and learning goals. Our work consists in making a comparative study of its models at the functional level and characteristics in the educational adaptive hypermedia systems in order to conclude the most effective learner model that can help to improve the academic results and to make online learning more accessible and effective for all learners, the results prove that the model based on learning styles allows a better adaptation in the educational adaptive hypermedia systems, its development will be devoted to the work of the next article.

A comparative study of characteristics and preferences to learner models in educational adaptive hypermedia systems

A comparative study of characteristics and preferences to learner models in educational adaptive hypermedia systems

Forecasting Graduation Schedule Model of Higher Education Learners using Feature Selection Techniques

Forecasting Graduation Schedule Model of Higher Education Learners using Feature Selection Techniques

Prediction and Evolution of Distance Education Learners’ Feedback Attitudes by a Deep Learning Approach

In recent years, the feedback attitude towards teachers has become one of the most important factors affecting learners' trust in teachers' teaching ability and changing learning plans, so the research on learners' feedback attitude has received extensive attention from experts and scholars at home and abroad. Most of the existing relevant research focuses on empirical research and behavioral research using questionnaires and scales from a theoretical point of view, which need to be further verified in terms of reliability and scientificity. Therefore, this paper conducts a research on the prediction and evolution of distance education learners’ feedback attitudes towards teachers based on deep learning. With the help of deep learning technology, it is easy to discover the advantages of distributed feature representation of data, and ARIMA model is combined with BP neural network model to construct a predictive model of distance education learners' feedback attitudes towards teachers. This paper makes the evolution analysis of distance education learners' feedback attitudes towards teachers, and introduces the assumptions and principles of the evolutionary analysis model in detail. The experimental results verify the effectiveness of the proposed model, and the analysis results of learner attitude evolution based on distance English learning as an example are given.

The Affordances of Multivariate Elo-Based Learner Modeling in Game-Based Assessment

Previous research and experiences have indicated the potential that games have in educational settings. One of the possible uses of games in education is as game-based assessments (GBA), using game tasks to generate evidence about skills and content knowledge that can be valuable. There are different approaches in the literature to implement the assessment machinery of these GBA, all of them having strengths and drawbacks. In this paper, we propose using multivariate Elo-based learner modeling, as we believe it has a strong potential in the context of GBA for three aims: 1) to simultaneously measure students competence across several knowledge components in a game, 2) to predict task performance, and 3) to estimate task difficulty within the game. To do so, we present our GBA Shadowspect, which is focused on solving geometry puzzles, and we depict our implementation using data collected from several high schools across the USA. We obtain high performing results (AUC of 0.87) and demonstrate that the model enables analysis of how each student's competency evolves after each puzzle attempt. Moreover, the model provides accurate estimations of each task's difficulty, enabling iterative improvement of the game design. This study highlights the potential that multivariate Elo-based learner modeling has within the context of GBA, sharing lessons learned, and encouraging future researchers in the field to consider this algorithm to build their assessment machinery.

When time matters: Mechanisms of change in a mediational model of foreign language playfulness and L2 learners’ emotions using latent change score mediation model

In a dynamic system, time-dependent links between affective factors can provide more information than the level of response within a single isolated system. In the present study, influenced by the positive psychology movement and the complex dynamic systems theory in the domain of second language acquisition, first, we dealt with change in terms of short-term dynamics and long-term trajectories of foreign language enjoyment (FLE), foreign language boredom (FLB), and foreign language playfulness (FLP) in a sample of 636 learners of English as a foreign language (EFL) using univariant latent change score (LCS) models. Then, we explored the developmental processes involved in how changes in FLE and FLP were associated with changes in FLB. In particular, we tested mediation models to see whether the growth of FLP acts as a mediator between FLE and FLB changes in a multivariant LCS mediation (LCSM) model. The findings showed that (a) in a multivariant LCS model, FLE and FLP increases independently predicted decreases in FLB over time and (b) the growth of FLP acted as a mediator between variation in FLE and FLB. Participants showed interindividual and intraindividual divergences in their L2 emotions, not just on the first time of measurement, but also in short-term dynamics and long-term trajectories. The findings facilitate understanding of the complicated mechanism of variation in L2 emotions, thus potentially contributing to enhancement of pedagogical practices and learning outcomes.

Orthogonal stochastic configuration networks with adaptive construction parameter for data analytics

As a randomized learner model, SCNs are remarkable that the random weights and biases are assigned employing a supervisory mechanism to ensure universal approximation and fast learning. However, the randomness makes SCNs more likely to generate approximate linear correlative nodes that are redundant and low quality, thereby resulting in non-compact network structure. In light of a fundamental principle in machine learning, that is, a model with fewer parameters holds improved generalization. This paper proposes orthogonal SCN, termed OSCN, to filtrate out the low-quality hidden nodes for network structure reduction by incorporating Gram–Schmidt orthogonalization technology. The universal approximation property of OSCN and an adaptive setting for the key construction parameters have been presented in details. In addition, an incremental updating scheme is developed to dynamically determine the output weights, contributing to improved computational efficiency. Finally, experimental results on two numerical examples and several real-world regression and classification datasets substantiate the effectiveness and feasibility of the proposed approach.

Clustering children's learning behaviour to identify self-regulated learning support needs

When children are learning using adaptive learning technologies (ALTs), the technology builds a learner model, which creates temporal trajectories providing insight into how children's knowledge develops. Based on this learner model, ALTs adjust the difficulty of problems for each child, yet children still need to regulate their practice behaviour and uphold effort and accuracy. The temporal trajectories are consequently likely to, besides showing children's knowledge development, be indicative of children's regulation. Therefore, we explore clusters of these trajectories to further identify failure in children's self-regulated learning (SRL) and potential support needs. We propose a data-driven approach to cluster 354 trajectories of 134 5th graders learning three skills with different complexity. The resulting 9 clusters were interpreted using practice accuracy and effort as indicators of regulation of practice behaviour and prior and post-knowledge and learning gain as indicators of knowledge development. The differences between clusters regarding these indicators signal there are different levels of SRL failure and, consequently, different SRL support needs: high accuracy and knowledge development indicate minimal support needs, whereas clusters with low accuracy, showing no knowledge development, indicate extensive SRL support needs. In conclusion: clusters of temporal patterns in children's learning data can identify SRL support is needed.

Differentiation Models for the Curriculum of Gifted and Talented Individuals: A Literature Review

The purpose of this research is to inform about the differentiation models related to the teaching programs of the gifted and talented individuals in the scope of the special education and to prove the necessity of using these models with related researches. Differentiation, which is accepted as changing the education process according to the individual differences of the students in the education of the gifted, is used as a tool in enrich the education programs of these individuals. In this study, information about gifted is given, the necessity of differentiating the curriculum of these individuals is emphasized and information about differentiation models is given. In this context, the definitions of The Maker Curriculum Differentiation Model, The Parallel Curriculum Model, The Curriculum Narrowing Model, The Integrated Curriculum Model, The Grid Curriculum Model, The Three-Stage Purdue Model, The Enrichment Triad Model, The Autonomous Learner Models about the curriculum of gifted students and supporting studies in the context are researched.

Evaluation and guide for embedding opioid use disorder education in health professions’ curricula

BackgroundMorbidity and mortality from Opioid Use Disorder is a health crisis in the United States. During the COVID-19 pandemic, there was a devastating increase of 38.4% in overdose deaths from the 12-month period leading up to June 2019 compared with the 12-month period leading up to May 2020, primarily driven by synthetic opioids. Buprenorphine is an effective medication for opioid use disorder but uptake is slow due in part to lack of provider knowledge, confidence, and negative attitudes/stigma toward patients with OUD. Addressing these barriers in academic training is a promising approach to building workforce able to effectively treat opioid use disorder.MethodsOur university developed a training for pre-licensure physicians, physician assistants and psychiatric nurse practitioners that included the DATA Waiver training and a shadowing experience. Expected outcomes included improved knowledge, skills and attitudes about persons with OUD and buprenorphine treatment, plans to provide this treatment post-graduation, for pre-licensure learners to have completed all requirements to prescribe buprenorphine post-graduation, and for the training to be embedded into school’s curricula.ResultsResults were positive overall including improved knowledge and attitudes toward persons with OUD, better understanding of the benefits of this treatment for patients, increased confidence and motivation to provide this treatment post-graduation. The training is now embedded in each program’s graduation requirements.ConclusionDeveloping a didactic and experiential training on buprenorphine treatment for opioid use disorder and embedding it into medical, physician assistant, and psychiatric nurse practitioner licensure programs can help prepare future providers to treat opioid use disorder in a range of settings. Key to replicating this program in other university settings is to engage faculty members who actively provide treatment to persons with OUD to ensure shadowing opportunities and serve as role models for learners.

Semantic surprise predicts the N400 brain potential

Language is central to human life; however, how our brains derive meaning from language is still not well understood. A commonly studied electrophysiological measure of on-line meaning related processing is the N400 component, the computational basis of which is still actively debated. Here, we test one of the recently proposed, computationally explicit hypotheses on the N400 – namely, that it reflects surprise with respect to a probabilistic representation of the semantic features of the current stimulus in a given context. We devise a Bayesian sequential learner model to derive trial-by-trial semantic surprise in a semantic oddball like roving paradigm experiment, where single nouns from different semantic categories are presented in sequences. Using experimental data from 40 subjects, we show that model-derived semantic surprise significantly predicts the N400 amplitude, substantially outperforming a non-probabilistic baseline model. Investigating the temporal signature of the effect, we find that the effect of semantic surprise on the EEG is restricted to the time window of the N400. Moreover, comparing the topography of the semantic surprise effect to a conventional ERP analysis of predicted vs. unpredicted words, we find that the semantic surprise closely replicates the N400 topography. Our results make a strong case for the role of probabilistic semantic representations in eliciting the N400, and in language comprehension in general.

An automatic identification method of imbalanced lithology based on Deep Forest and K-means SMOTE

Accurate identification of lithology is an important basis for oil and gas exploration and reservoir geological evaluation. Logging parameters often have a complex nonlinear relationship with lithology. With the development of artificial intelligence technology, a variety of data mining algorithms have been applied to lithology identification with logging data. However, due to the constraints of practical conditions, the number of labeled lithology samples is small and the imbalance between classes is obvious, which usually affect the results of lithology identification. In this paper, we proposed a hybrid unbalanced lithology identification method based on Deep Forest and K-means SMOTE to solve the above mentioned problems. Deep Forest is the first deep model of a non-differential form base learner which can perform layer-by-layer processing and feature enhancement. Compared with traditional shallow model, ensemble model and deep neural networks, it has the best performance with the accuracy of 96.17%. K-means SMOTE oversampling is able to filter subclusters and balance the dataset by oversampling only within safe regions, reducing interference from noise points and boundary ambiguity. We achieved further performance improvements in minority lithology classification via K-means SMOTE oversampling with the accuracy of 97.17%. The comprehensive evaluation results showed that the method proposed in this paper has a good effect on the identification of unbalanced lithology, and has a practical application prospect.