Learning Criterion Research Articles

Multiple fairness criteria in decision tree learning

The use of algorithmic decision-making systems based on machine learning models has led to a need for fair (unbiased) and explainable classification outcomes. In particular, machine learning algorithms can encode biases, which might result in discriminatory decisions for certain groups such as gender, race, or age. Although a number of works on decision tree learning have been proposed to decrease the chance of discrimination, they usually focus on the use of a single fairness metric. In general, creating a model based on a single fairness metric is not a sufficient way to mitigate discrimination since bias can originate from various sources—e.g., the data itself or the optimization process. In this paper, we propose a novel decision tree learning process that utilizes multiple fairness metrics to address both group and individual discrimination. This is achieved by extending the attribute selection procedure to consider not only information gain but also gain in fairness. Computational experiments on fourteen different datasets with various sensitive features demonstrate that the proposed Fair-C4.5 models improve fairness without a loss in predictive accuracy when compared to the well-known C4.5 and the fairness-aware FFTree algorithms.

BALANCED SCORECARD DESIGN AS A MODEL FOR MEASUREMENT OF PUSKESMAS PERFORMANCE IN EAST LOMBOK DISTRICT

The objective of this study is to develop a performance measuring model based on the balanced scorecard framework at the UPT Puskesmas Suela in East Lombok Regency, using the Analytic Hierarchy Process (AHP) technique. This research employed qualitative approaches. Data were gathered with a paired comparison questionnaire employing the Analytic Hierarchy Process (AHP) scale. The performance evaluation of community health centers is conducted using four reference standards, including the financial perspective, community perspective, internal business process perspective, and development and learning perspective. Utilizing this technique for data processing is beneficial in making judgments based on many factors. The research findings indicate that the primary criterion of utmost importance for community health centers is finance, with a weight of 0.465450. The subsequent prioritization criterion is society, assigned a weight of 0.214555. Additionally, both the internal business process criteria and development and learning criteria hold equal significance, each with a weight of 0.112321. Related indicates that Puskesmas as a public sector agency focuses on improving performance on financial perspectives.

Improving Understanding of Data Processing Concepts through the TARL Approach (Teaching at the Right Level) in Class 5 Students of SD Muhammadiyah 23 Surakarta

This study aims to enhance the understanding of data processing concepts among fifth-grade students at SD Muhammadiyah 23 Surakarta through the Teaching at the Right Level (TaRL) approach. The research was conducted using a Classroom Action Research (CAR) methodology, which involved two cycles, each consisting of two meetings. The primary issue identified was the low level of students' comprehension of data processing concepts, such as data collection, classification, interpretation, and the use of diagrams and tables. To address this, an active and contextual learning approach was implemented, involving real-life data collection and analysis, the use of digital tools for data visualization, and Problem-Based Learning (PBL) strategies.In the first cycle, the intervention revealed limited improvement in student engagement and learning outcomes. Consequently, the second cycle introduced revised group formations and additional instructional strategies, including ice-breaking activities. The results showed a significant improvement in student participation and understanding, with an increase in average class scores from 65 in the first cycle to 81 in the second cycle. The percentage of students meeting the learning criteria rose from 14.28% to 82.14%. This study concludes that applying TaRL, along with improved group dynamics and interactive learning methods, effectively enhances students' data processing skills and overall learning outcomes.

Effectiveness of Three Vocabulary Learning Tasks from The Perspective of Technique Feature Analysis

This study investigated the effectiveness of vocabulary learning tasks based on the Technique Feature Analysis framework. The framework provides criteria for evaluating the effectiveness of vocabulary learning tasks. The study investigated the correlation between the predictability of the framework and the task’s effect on vocabulary learning and retention. The study employed an experimental design in which participants were assigned to three different vocabulary tasks involving 15 target words. The tasks were designed to vary in the presence of the vocabulary learning criteria outlined in the Technique Feature Analysis. Learners’ knowledge of word forms and word meanings was assessed immediately and one week later. The results provided mixed support for the Technique Feature Analysis framework. While no significant differences were found in immediate form-meaning recognition, and delayed form recognition, the framework was predictive of learners’ delayed meaning recognition performance.

Trajectory Recognition and Working Condition Analysis of Rod Pumping Systems Based on Pose Estimation Method with Heatmap-Free Joint Detection

Summary Rod pump systems are the primary production tools in oilfield development. Analyzing their working conditions provides a theoretical foundation for formulating production optimization plans and adjusting equipment parameters. Existing machine learning–based condition analysis methods rely on dynamometer cards and cannot capture the actual operational trajectory of the pumping unit. To address this issue, this paper proposes a keypoint detection method for pumping units based on pose estimation of heatmap-free joint detection from video data. A data annotation scheme suitable for the task of detecting pumping unit keypoints was developed, and the learning criteria for this task were optimized. An end-to-end heatmap-free pose estimation algorithm was used to process images of the pumping unit, yielding predicted keypoint positions, thereby enabling the identification of the keypoint motion trajectories of the pumping unit. Experiments validated the proposed method and compared it with general learning criteria. Results show that this method accurately captures the keypoint positions of the pumping unit, with the optimized learning criteria model improving by more than 5% compared with general methods and increasing the keypoint object keypoint similarity (OKS) by more than 30%. The model’s results can be used for the actual operational trajectory recognition of the pumping unit, automatically calculating the motion parameters of the polished rod, and intelligently assessing the balance and working condition analysis of the pumping unit. This realizes the intelligent application of video surveillance data, significantly contributing to the dynamic study of rod pump systems.

Serial visual reversal learning in captive black-handed spider monkeys, Ateles geoffroyi

Recent research suggests that socio-ecological factors such as dietary specialization and social complexity may be drivers of advanced cognitive skills among primates. Therefore, we assessed the ability of 12 black-handed spider monkeys (Ateles geoffroyi), a highly frugivorous platyrrhine primate with strong fission-fusion dynamics, to succeed in a serial visual reversal learning task. Using a two-alternative choice paradigm we first trained the animals to reliably choose a rewarded visual stimulus over a non-rewarded one. Upon reaching a pre-set learning criterion we then switched the reward values of the two stimuli and assessed if and how quickly the animals learned to reverse their choices, again to a pre-set learning criterion. This stimulus reversal procedure was then continued for a total of 80 sessions of 10 trials each. We found that the spider monkeys quickly learned to reliably discriminate between two simultaneously presented visual stimuli, that they succeeded in a visual reversal learning task, and that they displayed an increase in learning speed across consecutive reversals, suggesting that they are capable of serial reversal learning-set formation with visual cues. The fastest-learning individual completed five reversals within the 80 sessions. The spider monkeys outperformed most other primate and nonprimate mammal species tested so far on this type of cognitive task, including chimpanzees, with regard to their learning speed in both the initial learning task and in the first reversal task, suggesting a high degree of behavioral flexibility and inhibitory control. Our findings support the notion that socio-ecological factors such as dietary specialization and social complexity foster advanced cognitive skills in primates.

Large-Scale Satisfaction Rating-Driven Selection of New Energy Vehicles: A Basic Uncertain Linguistic Information Bonferroni Mean-Based MCGDM Approach Considering Criteria Interaction

The continuous revolution of new energy technologies and the introduction of subsidy policies have promoted green consumers’ willingness to purchase new energy vehicles and automotive online service platforms have disclosed vehicle reputation and consumer satisfaction ratings information. However, due to issues such as uncertain data quality, large data volumes, and the emergence of positive reviews, the cost for potential car buyers to acquire useful decision-making knowledge has increased. Therefore, it is necessary to develop a scientific decision-making method that leverages the advantages of large-scale consumer satisfaction ratings to support potential car buyers in efficiently acquiring credible decision-making knowledge. In this context, the Bonferroni mean (BM) is a prominent operator for aggregating associated attribute information, while basic uncertain linguistic information (BULI) represents both information and its credibility in an integrated manner. This study proposes an embedded-criteria association learning BM operator tailored to large-scale consumer satisfaction ratings-driven scenarios and extends it to the BULI environment to address online ratings aggregation problems. Firstly, to overcome the limitations of BM with weighted interaction (WIBM) when dealing with independent criteria, we introduce an adjusted WIBM operator and extend it to the BULI environment as the BULIWIBM operator. We discuss fundamental properties such as idempotence, monotonicity, boundedness, and degeneracy. Secondly, addressing the constraints on interaction coefficients in BM due to subjective settings, we leverage expert knowledge to explore potential temporal characteristics hidden within large-scale consumer satisfaction ratings and develop a method for learning criteria and interaction coefficients. Finally, we propose a conversion method between user credibility-based ratings and BULI. By combining this method with the proposed adjusted BM operator, we construct a multi-criteria group decision-making (MCGDM) approach for product ranking driven by large-scale consumer satisfaction ratings. The effectiveness and scientific rigor of our proposed methods are demonstrated through solving a new energy vehicle selection problem on an online service platform and conducting comparative analysis. The case analysis and comparative analysis results demonstrate that the interaction coefficients, derived from expert knowledge and 42,520 user ratings, respectively, fell within the ranges of [0.2391, 0.7857] and [0.6546, 1.0]. The comprehensive interaction coefficient lay within the range of [0.4674, 0.7965], effectively mitigating any potential biases caused by subjective or objective factors. In comparison to online service platforms, our approach excels in distinguishing between alternative vehicles and significantly impacts their ranking based on credibility considerations.

The Use of Scavenger Hunt Game to Promote an English Learning Atmosphere For Young Learners

The learning atmosphere is one aspect that can support success in the learning process. However, to create a good and enjoyable learning atmosphere, teachers must use a technique or strategy that is suitable, one of which is the use of the Scavenger Hunt game. This research aimed to describe the use of a Scavenger Hunt game in promoting the English learning atmosphere which is conducted at MIS Mambaul Islam Soko Tuban. To explain the extent of the Scavenger Hunt game in promoting the English learning atmosphere, the researcher employed a case-study research approach to get more information in detail. The data were collected through the observation and questionnaire. The population of this research was the fifth-grade students of MIS Mambaul Islam Soko Tuban in the academic year 2023/2024 which consisted of 24 students. The result showed that the use of the Scavenger Hunt game was effective in promoting the English learning atmosphere evidenced by the fulfillment of good and positive Atmosphere learning criteria after the implementation of the Scavenger Hunt game.

Understanding validity criteria in technology-enhanced learning: A systematic literature review

Technological aids are ubiquitous in today's educational environments. Whereas much of the dust has settled in the debate on how to validate traditional educational solutions, in the area of technology-enhanced learning (TEL) many questions still remain. Technologies often abstract away student behaviour by condensing actions into numbers, meaning teachers have to assess student data rather than observing students directly. With the rapid adoption of artificial intelligence in education, it is timely to obtain a clear image of the landscape of validity criteria relevant to TEL. In this paper, we conduct a systematic review of research on TEL interventions, where we combine active learning for title and abstract screening with a backward snowballing phase. We extract information on the validity criteria used to evaluate TEL solutions, along with the methods employed to measure these criteria. By combining data on the research methods (qualitative versus quantitative) and knowledge source (theory versus practice) used to inform validity criteria, we ground our results epistemologically. We find that validity criteria tend to be assessed more positively when quantitative methods are used and that validation framework usage is both rare and fragmented. Yet, we also find that the prevalence of different validity criteria and the research methods used to assess them are relatively stable over time, implying that a strong foundation exists to design holistic validation frameworks with the potential to become commonplace in TEL research.

Development of Website based Multiple Representation on Buffer Solution Topic in Chemistry Learning

Buffer solution is one of chemistry difficult topics. It contains theory, calculation, and experimentation, so it requires good analysis and understanding of concepts. The results of interviews with students in Bintan Regency showed that the sub material of determining buffer solution components and analyzing the working principles of buffer solutions was the most difficult part because it was abstract. Multiple representations on the website media will visualize abstract things from the topic of buffer solution so as to increase the understanding of concepts from students. The research aims to produce multiple representation website on buffer solution topics and to determine its feasibility, practicality, and readability. The research method used is research and development (RD) following the 4D development model by Thiagrajan (define, design, develop, and disseminate). The sample research used 21 students and five chemistry teachers in Bintan Regency to collect user responses related to practicality and readability. The data analysis calculates the average score of the validation and response values and classifies them into the available categories. The results of the validation show that the average score of material experts is 63.5 (learning, materials, multiple representations, and evaluation aspects), while the average score of media experts is 71.5 (graphic design, software engineering, and media selection criteria aspects), which are classified as very good. The results of practicality and readability responses are 119 (learning, materials, multiple representations, evaluation, graphic design, software engineering, and media selection criteria) and 70.5 (learning, graphic design, and software engineering), which are very good categories

A machine learning decision criterion for reducing scan time for hyperspectral neutron computed tomography systems

We present the first machine learning-based autonomous hyperspectral neutron computed tomography experiment performed at the Spallation Neutron Source. Hyperspectral neutron computed tomography allows the characterization of samples by enabling the reconstruction of crystallographic information and elemental/isotopic composition of objects relevant to materials science. High quality reconstructions using traditional algorithms such as the filtered back projection require a high signal-to-noise ratio across a wide wavelength range combined with a large number of projections. This results in scan times of several days to acquire hundreds of hyperspectral projections, during which end users have minimal feedback. To address these challenges, a golden ratio scanning protocol combined with model-based image reconstruction algorithms have been proposed. This novel approach enables high quality real-time reconstructions from streaming experimental data, thus providing feedback to users, while requiring fewer yet a fixed number of projections compared to the filtered back projection method. In this paper, we propose a novel machine learning criterion that can terminate a streaming neutron tomography scan once sufficient information is obtained based on the current set of measurements. Our decision criterion uses a quality score which combines a reference-free image quality metric computed using a pre-trained deep neural network with a metric that measures differences between consecutive reconstructions. The results show that our method can reduce the measurement time by approximately a factor of five compared to a baseline method based on filtered back projection for the samples we studied while automatically terminating the scans.

Graph Embedded Ensemble Deep Randomized Network for Diagnosis of Alzheimer's Disease.

Randomized shallow/deep neural networks with closed form solution avoid the shortcomings that exist in the back propagation (BP) based trained neural networks. Ensemble deep random vector functional link (edRVFL) network utilize the strength of two growing fields, i.e., deep learning and ensemble learning. However, edRVFL model doesn't consider the geometrical relationship of the data while calculating the final output parameters corresponding to each layer considered as base model. In the literature, graph embedded frameworks have been successfully used to describe the geometrical relationship within data. In this paper, we propose an extended graph embedded RVFL (EGERVFL) model that, unlike standard RVFL, employs both intrinsic and penalty subspace learning (SL) criteria under the graph embedded framework in its optimization process to calculate the model's output parameters. The proposed shallow EGERVFL model has only single hidden layer and hence, has less representation learning. Therefore, we further develop an ensemble deep EGERVFL (edEGERVFL) model that can be considered a variant of edRVFL model. Unlike edRVFL, the proposed edEGERVFL model solves graph embedded based optimization problem in each layer and hence, has better generalization performance than edRVFL model. We evaluated the proposed approaches for the diagnosis of Alzheimer's disease and furthermore on UCI datasets. The experimental results demonstrate that the proposed models perform better than baseline models. The source code of the proposed models is available at https://github.com/mtanveer1/.

Are lateralized and bold fish optimistic or pessimistic?

Cognitive bias is defined as the influence of emotions on cognitive processes. The concept of the cognitive judgement bias has its origins in human psychology but has been applied to animals over the past 2 decades. In this study we were interested in determining if laterality and personality traits, which are known to influence learning style, might also be correlated with a cognitive bias in the three-spined sticklebacks (Gasterosteus aculeatus). We used the judgement bias test with the go/no-go procedure where fish were first trained to discriminate between a black and white card and, after reaching a minimum learning criterion, tested their response to an ambiguous card (grey). Optimistic subjects were expected to have a high expectation of reward associated with an ambiguous stimulus, whereas pessimistic subjects a high expectation of non-reward. We used an emergence and a mirror test to quantify boldness and laterality, respectively. We hypothesised that male, bolder and more strongly lateralized fish would be more optimistic than female, shy and less strongly lateralised fish. We found that males and more strongly lateralized fish were more optimistic than females and less strongly lateralized fish. In addition, bold males were more optimistic than shy males as we predicted, but females showed the opposite pattern. Finally, fish trained on the black colour card learned the training task faster than those trained on a white card. Our results indicate that both laterality and personality traits are linked to animals’ internal states (pessimistic or optimistic outlooks) which likely has broad implications for understanding animal behaviour particularly in a welfare context.

Optimal design of experiments for computing the fatigue life of an offshore wind turbine based on stepwise uncertainty reduction

The design of an offshore wind turbine to resist fatigue damage during its whole service life requires to estimate an expectation over the pluri-annual joint statistics of wind and wave variables. Using a full factorial-based integration for the estimation of the cumulative fatigue damage represents a tremendous computational cost with aero-servo-hydro-elastic solvers which is generally not affordable by industrial designers. To overcome this limitation, strong approximations with lumping of environmental discretized joint probability (scatter diagram) are generally employed. We present in this paper a new method, called MAKSUR, involving the iterative enrichment of a design of experiments tailored to provide a good approximation of the long term mean damage. This method relies on a Kriging response surface with a learning criterion defined as the variance of the mean damage integral. It is compared to another previous similar approach called AK-DA, also dedicated to damage prediction, but is shown to converge more efficiently and with less numerical parameters to define by the user. The potential of the method for offshore wind turbine is demonstrated by a realistic 6D floating wind turbine case study with six wind and wave input variables.

Strain and age but not maternal feeding of n-3 fatty acids affect the performance of laying hen offspring in a series of associative learning tasks

Commercial poultry are typically fed diets with fatty acid compositions that differ considerably from wild-type diets in the omega-6 to omega-3 FA ratio (n-6: n-3). Maternal dietary n-3 FA have been shown to affect offspring brain development and cognition in various avian and mammalian species. This study explored the effects of feeding egg-type layer breeders with flaxseed on the performance of their female offspring in a three-stage discrimination task. Shaver White and ISA Brown breeders were fed either a control diet (n-6: n-3 = 14.7:1) or a flaxseed-enriched diet (n-6: n-3 = 5.3:1) throughout the rearing and laying phases. At 30 and 36 weeks of age, eggs were collected, incubated and hatched. Offspring were housed in enriched floor pens and received a control diet. At 12 (pullets) and 33 (hens) weeks of age, three learning tasks were conducted: behavioural shaping (N=437), discrimination test (N=350), and reversal learning (N=213). Food rewards (sweetcorn and raisins) were placed in the “reward” cup. Behavioural shaping consisted of five phases (P): P1 clear cup, P2 coloured cup, P3 coloured cup half-covered with a lid, P4 coloured cup three quarter covered with a lid, and P5 coloured cup covered with a lid. One cup, alternating between blue and green, was presented. Chickens had to eat the food reward from the cup in 4/5 trials to pass each phase. Two cups (green and blue) were presented during discrimination testing and reversal learning with food only in the predetermined “reward” cup. Learning criterion was choosing the correct cup in 4/5 trials for three consecutive days in discrimination testing and for two consecutive days in reversal learning. Data were analysed using survival class analyses. The hazard ratio (HR), 95% confidence intervals (CI) [LL, UL] and p-values to determine the probability of reaching learning criteria are reported. In P1 of behavioural shaping, Brown hens were likelier to reach criteria (HR=2.28, 95% CI [1.48, 3.55], p<0.001). For overall success in behavioural shaping (P1–5), hens were likelier to reach criteria than pullets (HR=2.42, 95% CI [1.50, 3.91], p<0.001). White chickens were likelier to reach criteria in the discrimination test than Brown chickens (HR=1.03, 95% CI [1.01, 1.75], p=0.04). Finally, pullets were likelier to reach criteria in reversal learning than hens (HR=0.44, 95% CI [0.23, 0.86], p=0.016). These results show that age, strain, and their interaction, but not maternal n-3 FA diet, affected the success of chickens in learning tasks.

Effects of an Augmentative and Alternative Communication Intervention Package on Socio-Communicative Behaviors Between Minimally Speaking Autistic Children and Their Peers.

The purpose of this study is to investigate the effectiveness of an augmentative and alternative communication (AAC) intervention package consisting of systematic instruction and aided modeling with speech-output technologies on the acquisition, maintenance, and generalization of socio-communicative behaviors-initiating a request for a turn, answering questions, and commenting-in four, minimally speaking (MS) autistic children between the ages of 6 and 9 years. A multiple-probe design across behaviors replicated across participants was implemented to evaluate the effects of systematic instruction and aided modeling on initiating requests for a turn, answering questions, and commenting behaviors. Additionally, a pre- and posttreatment multiple-generalization-probes design was used to assess generalization across peers. Visual analyses demonstrated experimental control for two participants (i.e., Derek, Ajay) showing a functional relationship between the intervention and outcomes across all social communicative behavior. For one participant (i.e., Matthew), experimental control could not be established because he did not reach the learning criterion for commenting. The fourth participant (i.e., John) transferred to a different school after making some progress on requesting. Effect size indicator analyses corroborated these findings, indicating medium-to-strong effects for initiating requests for a turn strong effects for answering questions, and medium-to-strong effects for commenting. Generalization of socio-communicative behaviors from researcher to a typically developing peer was variable across participants. Participants maintained socio-communicative behaviors 3 weeks after the last intervention session with varying degrees of success. The outcomes of this study suggest that aided modeling and systematic instruction using speech-output technologies may lead to gains in socio-communicative behaviors in some MS autistic children. https://doi.org/10.23641/asha.25799935.

Perceived Utility and Learning by Dominican University Students in Virtual Teaching–Learning Environments: An Analysis of Multiple Serial Mediation Based on the Extended Technology Acceptance Model

The global pandemic caused by the SARS-CoV-2 virus brought about a true revolution in the predominant teaching–learning processes (i.e., face-to-face environment) that had been implemented up to that point. In this regard, virtual teaching–learning environments (VTLEs) have gained unprecedented significance. The main objectives of our research were to define an explanatory theoretical model and to test a multiple serial mediation model with four variables in series (one independent variable plus three mediators) to relate perceived utility (independent variable) in the use of a VTLE and perceived learning (dependent or criterion variable) in such contexts, taking into account the mediation of subjective norm (mediator 1), ease of use (mediator 2), and intention to use behavior (mediator 3), and using the extended technology acceptance model as the theoretical framework. Additionally, we aimed to analyze the direct and indirect relationships and effects among the variables that constituted the proposed model. Methodologically, the research can be classified as a cross-sectional causal ex post facto design. A representative sample of students enrolled in higher education institutions in the Dominican Republic was used as the research population, and a standardized Likert scale was administered to measure the five dimensions of the proposed model. Finally, it is worth noting that the obtained results indicate that all direct and indirect effects considered in the model were statistically significant, except for the indirect effect, where the four predictor variables were arranged in series to verify their influence on the criterion variable: perceived learning.

The Most Ideal L2 Teaching Method

Though it is possible to see numerous researches on the so-called language teaching methods, there is still a pressing need to study them comparatively and from different angles. The article studies some specific methods with regard to covering language skills, mother tongue usage, usage of cutting-edge technological devices, comprehensibility gain. Besides, there is a strong focus on the conduction of the lesson regarding whether the lesson is student-centered, teacher-centered or self-centered. The aim of the article is to reveal the best language learning method combining the positive features of all language learning criteria. In the study, 20 local and foreign non-English speaking students with B1 level were involved in the experiment. They were divided into four groups which learnt English through a specific method. Most of the participants were locals except 6 who were of Turkish and Iranian origins. Speaking skill of the Iranian students was relatively better in comparison with others. At the end of the experiment, the comprehensibility gain of each group was graded accordingly. The Uniskills method proved to be the most effective and qualitative method compared to other methods in terms of coverage of language skills. The second effective method was CLT which also encompassed all four skills. It was revealed that the usage of mother tongue in English language learning is better than not using it. Since language input makes progress while using native language. Comparing the source and target languages makes a clear picture of language structure in the brain. Operation function is activated and cognitive skills develop significantly. It turned out that the usage of mother tongue in language learning is more effective and the students find the meanings of some unknown words obscure when they don’t use the native language. It became obvious that teacher-centered lesson is more efficient than the student-centered education.

Large-scale experiment in STEM education for high school students using artificial intelligence kit based on computer vision and Python

This study proposes an artificial intelligence (AI) kit for high school students in science, technology, engineering, and mathematics (STEM). The AI kit includes an edge AI machine and electronic components. A compact, purpose-built kit resembling a laptop was designed for ease of replication and portability. Utilizing pre-trained convolutional neural network models and computer vision algorithms, five Thai schools participated in on-site instructions. A quasi-experimental study assessed the students' learning outcomes using a paired sample t-test. Results revealed improved knowledge and reduced score variation. Additionally, gender analysis confirmed that both male and female students met the learning criteria. The students expressed satisfaction with the distinctive hardware and learning method employed during the class activities. Notably, the test results demonstrated that the AI kit enhanced students’ enthusiasm and facilitated comprehension.

Active Learning and Bayesian Optimization: A Unified Perspective to Learn with a Goal

Science and Engineering applications are typically associated with expensive optimization problem to identify optimal design solutions and states of the system of interest. Bayesian optimization and active learning compute surrogate models through efficient adaptive sampling schemes to assist and accelerate this search task toward a given optimization goal. Both those methodologies are driven by specific infill/learning criteria which quantify the utility with respect to the set goal of evaluating the objective function for unknown combinations of optimization variables. While the two fields have seen an exponential growth in popularity in the past decades, their dualism and synergy have received relatively little attention to date. This paper discusses and formalizes the synergy between Bayesian optimization and active learning as symbiotic adaptive sampling methodologies driven by common principles. In particular, we demonstrate this unified perspective through the formalization of the analogy between the Bayesian infill criteria and active learning criteria as driving principles of both the goal-driven procedures. To support our original perspective, we propose a general classification of adaptive sampling techniques to highlight similarities and differences between the vast families of adaptive sampling, active learning, and Bayesian optimization. Accordingly, the synergy is demonstrated mapping the Bayesian infill criteria with the active learning criteria, and is formalized for searches informed by both a single information source and multiple levels of fidelity. In addition, we provide guidelines to apply those learning criteria investigating the performance of different Bayesian schemes for a variety of benchmark problems to highlight benefits and limitations over mathematical properties that characterize real-world applications.