Learner Phase Research Articles

Point Cloud Registration Method Based on Improved TLBO for Landing Gear Components Measurement

When using point cloud technology to measure the dimension and geometric error of aircraft landing gear components, the point cloud data obtained after scanning may have certain differences because of the sophistication and diversity of the components that make up the landing gear. However, when using traditional point cloud registration algorithms, if the initial pose between point clouds is poor, it can lead to significant errors in the final registration results or even registration failure. Furthermore, the significant difference in registration results between point clouds can affect the final measurement results. Adopting Teaching-Learning-Based Optimization (TLBO) to solve some optimization problems has unique advantages such as high accuracy and good stability. This study integrates TLBO with point cloud registration. To increase the probability of using TLBO for point cloud registration to search for the global optimal solution, adaptive learning weights are first introduced during the learner phase of the basic TLBO. Secondly, an additional tutoring phase has been designed based on the symmetry and unimodality of the normal distribution to improve the accuracy of the solution results. In order to evaluate the performance of the proposed algorithm, it was first used to solve the CEC2017 test function. The comparison results with other metaheuristics showed that the improved TLBO has excellent comprehensive performance. Then, registration experiments were conducted using the open point cloud dataset and the landing gear point cloud dataset, respectively. The registration results showed that the point cloud registration method proposed in this paper has strong competitiveness.

A ranking improved teaching-learning-based optimization algorithm for parameters identification of photovoltaic models

Solar energy is an important clean energy source, primarily applied for photovoltaic (PV) power generation. The precise identification of PV system parameters is critical for system control and simulation, posing a challenge due to the models' non-linearity, implicitness, and multiple optimal properties. So, a ranking improved teaching-learning-based optimization (RITLBO) is developed in this work to solve the problem of identifying the parameters of the PV model. RITLBO is a meta-heuristic algorithm based on teaching-learning-based optimization (TLBO) that simulates classroom teacher-student interaction. In RITLBO, learners are classified into inferior and superior groups based on their fitness ranking. During the teacher phase, superior learners emulate the top three agents with the highest fitness for local search, while inferior learners engage in guided mutual learning for global search, effectively utilizing computing resources. In the learner phase, superior learners receive guided information, while inferior learners engage in broader information exchange, balancing exploration and exploitation. RITLBO and fourteen algorithms are used to identify the parameters for five different PV models to confirm that the RITLBO is effective. Statistical results and analysis demonstrate that RITLBO is accurate and reliable in identifying PV model parameters. RITLBO offers promising prospects in optimizing PV system parameters through its unique strategies.

Gaussian Mutation Based Teaching-Learning Optimization for reactive power dispatch

This paper develops the Gaussian Mutation Based Teaching-Learning Optimization (GMBTLO) to solve different reactive power dispatch problems. In GMBTLO, Gaussian random variables are instituted, which includes both the ‘Teacher phase’ as well as ‘Learner phase’ helps to improve the efficiency of searching. Simultaneously, it ensures more feasibility to obtain the global optimal. Gaussian random variables are presented in GMBTLO, where both the ‘Teacher phase’ as well as ‘Learner phase’ is considered. This advances searching competence and guarantee a great possibility of achieving global optimum without affecting the convergence speed. In this paper, GMBTLO is implemented to enhance solution quality as well as convergence speed. Developed GMBTLO is exploited for obtaining the control variables settings including terminal voltages of a generator, transformer taps, as well as output power of shunt VAR compensators (reactive) for achieving a reduced loss in real power transmission and enhanced voltage profile. GMBTLO is validated by taking IEEE 30-bus, 57-bus along with 118-bus test systems. Also, test outcomes are matched with the results already acquired from other implemented evolutionary techniques. It is concluded that the suggested GMBTLO extends a better solution.

An adaptive ranking teaching learning-based optimization algorithm to solve sensor deployment in harsh environments

Sensors are often deployed in harsh environments, in which some threats may endanger the safety of sensors. In this paper, a sensor deployment model is developed in Wireless Sensor Networks (WSNs), in which the coverage rate and the threat risk are considered simultaneously. The model is established as an optimization problem. An adaptive ranking teaching learning-based optimization algorithm (ARTLBO) is proposed to solve the problem. Learners are divided into inferior and superior groups. The teacher phase is boosted by replacing the teacher with the top three learners, and the learner phase is improved by providing some guidance for inferior learners. The experimental results show that the proposed ARTLBO algorithm can effectively optimize the model. The fitness values of the proposed model found by the proposed ARTLBO are 0.4894, 0.4886, which are better than its competitors. The algorithm can provide a higher coverage rate and lower threat risk.

Boosting constrained teaching–learning based optimization algorithm through group discussion

Teaching–learning based optimization algorithm (TLBO) is an efficient population-based nature-inspired meta-heuristic designed initially to solve unconstrained optimization problems. TLBO being a swam intelligence based algorithm is renowned for fast convergence. This characteristic, however, occasionally causes it to converge too soon to a non-globally optimal solution. Group discussion techniques are used in educational institutions all over the world to help participants to exchange ideas and information and to improve their skills. This work imitates the outlined strategy, therefore innovating TLBO. Here, a group of virtual students is selected corresponding to each student in the learner phase of TLBO to produce diversity in the population and to balance the exploration and exploitation capabilities of it, hence reducing the likelihood of early convergence. GTLBO stands for the new algorithm that incorporates group discussion. It is essential to incorporate some constraint handling techniques (CHTs) into the TLBO framework in order to handle constrained optimization problems (COPs). The most recent, effective, and enhanced version of the widely used CHT stochastic ranking (SR) is called hybrid stochastic ranking (HSR). HSR is added to the GTLBO structure as a selection operator for managing COPs; the resulting setup is referred to as HSR-GTLBO. Additionally, the suggested algorithm’s group size for discussion is sensitive, resulting in a number of constrained variants. The efficacy of these variations is assessed with the suit CEC 2017’s constrained problems. In conformity with the parameters settlement, comparison, and ranking criteria of CEC 2017, the simulation results of the proposed versions are compared and ranked using the most advanced state-of-the-art algorithms. As a result, in the competition, the newly developed HSR-GTLBO variants placed first, second, and third.

An Adaptive Two-Class Teaching-Learning-Based Optimization for Energy-Efficient Hybrid Flow Shop Scheduling Problems with Additional Resources

Energy-efficient scheduling problems with additional resources are seldom studied in hybrid flow shops. In this study, an energy-efficient hybrid flow shop scheduling problem (EHFSP) with additional resources is studied in which there is asymmetry in the machine. An adaptive two-class teaching-learning-based optimization (ATLBO) which has multiple teachers is proposed to simultaneously minimize the makespan and the total energy consumption. After two classes are formed, a teacher phase is first executed, which consists of teacher self-learning and teacher training. Then, an adaptive learner phase is presented, in which the quality of two classes is used to adaptively decide the learner phase or the reinforcement search of the temporary solution set. An adaptive formation of classes is also given. Extensive experiments were conducted and the computational results show that the new strategies are effective and that ATLBO was able to provide better results than comparative algorithms reported in the literature in at least 54 of 68 instances.

Offence trajectories of young Australian novice drivers

This paper aims to add to the sparse research on patterns of traffic offending in young novice drivers by examining the trajectories of young Australian novices throughout the graduated driving licensing system (GDLS). Data on licensing, crashes and traffic offences for all persons aged under 25 years first licensed from July 2007 to June 2008 in New South Wales were supplied by transport authorities, with follow-up to January 2014. Eight hierarchical categories of offence were defined by seriousness. Total numbers of offences per month were plotted for one year after initial licensure while under supervision (L), after transitions to the first (P1) and second (P2) provisional licence phases and to unrestricted licensure (U). After excluding novices not transiting to P1, 75,470 novices (51.5 % male, 77.7 % 16 years old at L) remained. Of these 63,070 had a year of follow-up after P2 (52.5 % male, 82.3 % 16 years old at L) and 28,861 had a year of follow-up after U (55.5 % males, 91.7 % 16 years old at L), Offences were uncommon under at L but spiked after P1 and again after P2. Omitting GDLS offences there was a small increase after U. Patterns were broadly similar by age at L and time in L. Novices aged at least 18 years at L had the highest offence rates in the learner phase; those aged 17 years at L had the highest rates in P1 and P2. The longer novices spent in the supervised driving phase the lower their subsequent offence rates even allowing for age at P1. Urban or regional residence had little influence on offence rates but lower aggregate socio-economic status as measured by a census-derived index was associated with appreciably higher offence rates. Speeding was the most common offence type; dangerous/negligent driving was common immediately after P1, declining thereafter. These results are in broad agreement with previous research in noting the increase in offending as GDLS restrictions eased. They highlight the need for better understanding of the function of a GDLS in improving the safety of young drivers. They may also inform future enforcement and education policies for novice drivers. For example, rehabilitative interventions should occur coupled with enforcement soon after citation. Visible display of GDLS status would assist their identification.

A teaching-learning-based optimization algorithm with reinforcement learning to address wind farm layout optimization problem

As the global demand for renewable energy continues to rise, wind energy has received widespread attention as an eco-friendly energy source. Wind power generation is regarded as one of the key means to reduce carbon emissions and achieve sustainable development. Usually, a mass of turbines works together to produce electricity in a wind farm. However, downstream turbines will inevitably be influenced by the wake generated by upstream turbines, resulting in unused wind energy being lost. To reduce the negative effects of the wake, maximization of wind farm output power, and minimization of wind farm cost, a teaching-learning-based optimization algorithm with reinforcement learning is proposed in this paper. The improvements of the proposed algorithm mainly include the following three points: i) the original serial structure of the algorithm is changed to a parallel structure to accelerate the convergence and improve the efficiency of the algorithm. ii) the parameter F, which is adjusted by RL, is proposed to adjust the selection of the updating phase due to the design of a parallel structure. iii) in the modified learner phase, an individual is added to participate in the update, and a selection probability is proposed to improve the ability of the algorithm to retain the information of superior individuals. To study the performance of the modified algorithm, it was first tested against 10 other advanced algorithms on a benchmark testing suite. They then ran numerical experiments on four hypothetical wind farm cases under two simulated wind conditions. Finally, the superiority of improved algorithm over others and the effectiveness of addressing wind farm layout problem are demonstrated by experimental results.

Optimizing Image Classification: Automated Deep Learning Architecture Crafting with Network and Learning Hyperparameter Tuning.

This study introduces ETLBOCBL-CNN, an automated approach for optimizing convolutional neural network (CNN) architectures to address classification tasks of varying complexities. ETLBOCBL-CNN employs an effective encoding scheme to optimize network and learning hyperparameters, enabling the discovery of innovative CNN structures. To enhance the search process, it incorporates a competency-based learning concept inspired by mixed-ability classrooms during the teacher phase. This categorizes learners into competency-based groups, guiding each learner's search process by utilizing the knowledge of the predominant peers, the teacher solution, and the population mean. This approach fosters diversity within the population and promotes the discovery of innovative network architectures. During the learner phase, ETLBOCBL-CNN integrates a stochastic peer interaction scheme that encourages collaborative learning among learners, enhancing the optimization of CNN architectures. To preserve valuable network information and promote long-term population quality improvement, ETLBOCBL-CNN introduces a tri-criterion selection scheme that considers fitness, diversity, and learners' improvement rates. The performance of ETLBOCBL-CNN is evaluated on nine different image datasets and compared to state-of-the-art methods. Notably, ELTLBOCBL-CNN achieves outstanding accuracies on various datasets, including MNIST (99.72%), MNIST-RD (96.67%), MNIST-RB (98.28%), MNIST-BI (97.22%), MNST-RD + BI (83.45%), Rectangles (99.99%), Rectangles-I (97.41%), Convex (98.35%), and MNIST-Fashion (93.70%). These results highlight the remarkable classification accuracy of ETLBOCBL-CNN, underscoring its potential for advancing smart device infrastructure development.

MTLBORKS-CNN: An Innovative Approach for Automated Convolutional Neural Network Design for Image Classification

Convolutional neural networks (CNNs) have excelled in artificial intelligence, particularly in image-related tasks such as classification and object recognition. However, manually designing CNN architectures demands significant domain expertise and involves time-consuming trial-and-error processes, along with substantial computational resources. To overcome this challenge, an automated network design method known as Modified Teaching-Learning-Based Optimization with Refined Knowledge Sharing (MTLBORKS-CNN) is introduced. It autonomously searches for optimal CNN architectures, achieving high classification performance on specific datasets without human intervention. MTLBORKS-CNN incorporates four key features. It employs an effective encoding scheme for various network hyperparameters, facilitating the search for innovative and valid network architectures. During the modified teacher phase, it leverages a social learning concept to calculate unique exemplars that effectively guide learners while preserving diversity. In the modified learner phase, self-learning and adaptive peer learning are incorporated to enhance knowledge acquisition of learners during CNN architecture optimization. Finally, MTLBORKS-CNN employs a dual-criterion selection scheme, considering both fitness and diversity, to determine the survival of learners in subsequent generations. MTLBORKS-CNN is rigorously evaluated across nine image datasets and compared with state-of-the-art methods. The results consistently demonstrate MTLBORKS-CNN’s superiority in terms of classification accuracy and network complexity, suggesting its potential for infrastructural development of smart devices.

An integrated hybrid metaheuristic model for the constrained scheduling problem

Several problems in the domains of project management (PM) and operations research (OP) can be classified as optimization problems which are classically NP-Hard. One such highly important problem is the resource constrained project scheduling problem (RCPSP). The main aim of this problem is to find a schedule of the lowest and optimum makespan to complete a project, which involves resource as well as precedence constraints. But, being classically NP-Hard, the RCPSP requires exponential computational resources as the problem complexity increases. Thus, approximate techniques like computational intelligence (CI) based approaches provide better chances of finding near optimal solutions. This paper presents the usage of a hybrid technique using the phases of teaching learning-based optimization (TLBO) metaheuristic integrated with operators like crossover and mutation from the genetic algorithm (GA). An integrated hybrid using TLBO and 2-point crossover is applied in the teacher and learner phases to the discrete RCPSP problem. Further, to diversify the population, and enhance global search, the mutation operator is applied. The proposed model is extensively tested on well-known benchmark test instances and has been compared with other seminal works. The encouraging results make evident the efficiency of the provided solution for the RCPSP problem of varying magnitudes.

Identification of photovoltaic module parameters by implementing a novel teaching learning based optimization with unique exemplar generation scheme (TLBO-UEGS)

The performance evaluation of a Photovoltaic (PV) system heavily relies on accurately estimating the parameters based on its current—voltage relationships. However, due to the PV model’s inherent complexity, obtaining these parameters with precision and efficiency is a challenging task. In this study, a new variant known as teaching learning-based optimization with unique exemplar generation schemes (TLBO-UEGS) is proposed to address PV module parameter estimation problems with robustness and effectiveness. To enhance the performance of TLBO-UEGS, a modified initialization scheme that leverages the strengths of chaotic maps and dynamic oppositional based learning is introduced. This scheme ensures the generation of an initial population with improved solution quality. Furthermore, both the modified teacher phase and modified learner phase are integrated within the TLBO-UEGS optimization framework. This integration allows for different learning strategies to be employed based on the fitness values of each learner, effectively updating their search trajectories. Within the modified teacher phase, two unique exemplar generation schemes are designed to facilitate more effective guidance for learners in the first half of the population while maintaining population diversity. Meanwhile, the modified learner phase emulates a realistic knowledge acquisition process by enabling learners in the second half of the population to engage in collaborative learning with multiple peer learners or retain valuable knowledge from previous learning processes. Extensive simulations demonstrate that TLBO-UEGS achieves superior results, with the minimum root mean square error (RMSE) values of 3.5644 × 10−04 ± 0.0014, 1.3237 × 10−04± 0.0043, and 6.6016 × 10−06± 0.00011 obtained for Photowatt-PWP201, Leibold Solar (LSM 20), and Leybold Solar (STE 4/100) PV modules, respectively.

The adoption of oncoplastic surgery: Is there a learning curve?

Oncoplastic surgery (OPS) is a form of breast conservation surgery involving partial mastectomy followed by volume displacement or replacement surgery. As the field of OPS is growing, we sought to determine if there was a learning curve to this surgery. A retrospective chart review was conducted of all patients who underwent OPS over a 6-year period with a single surgeon formally trained in both Plastic Surgery and Breast Oncology. Cumulative summation analysis (CUSUM) was performed on mean operative time to generate the learning curve and learning curve phases. Outcomes were compared between phases to determine significance. Mean operative time decreased significantly across the 6-year period, generating three distinct learning curve phases: Learner phase (cases 1-23), Competence phase (24-73), and Mastery phase (74 and greater). The overall positive margin rate was 10.9% and there was no significant difference in rates between phases (p = 0.49). Overall complication rates, reoperation rates, and locoregional recurrence remained the same across all phases (p = 0.16; p = 0.65; p = 0.41). The rate of partial nipple loss decreased between phases (p = 0.02). As with many complex operations, there does appear to be a learning curve with OPS, as the operative time and the rates of partial nipple loss decreased over time.

A multi-class teaching–learning-based optimization for multi-objective distributed hybrid flow shop scheduling

Distributed hybrid flow shop scheduling problem (DHFSP) has attracted some attention. In this study, DHFSP with sequence-dependent setup times is studied and a multi-class teaching–learning-based optimization (MTLBO) is proposed to minimize makespan and maximum tardiness simultaneously. A two-string representation is adopted. s classes are formed to improve search efficiency by implementing reward and punishment mechanism among them. Class evaluation is introduced and two teacher phases and one learner phase are applied in the evolution of each class. Elimination process acts on the worst class to avoid the waste of computing resource. A number of experiments are conducted and the computational results demonstrate that MTLBO is a very competitive method for DHFSP.

Optimizing boiler combustion parameters based on evolution teaching-learning-based optimization algorithm for reducing NOx emission concentration.

How to reduce a boiler's NOx emission concentration is an urgent problem for thermal power plants. Therefore, in this paper, we combine an evolution teaching-learning-based optimization algorithm with extreme learning machine to optimize a boiler's combustion parameters for reducing NOx emission concentration. Evolution teaching-learning-based optimization algorithm (ETLBO) is a variant of conventional teaching-learning-based optimization algorithm, which uses a chaotic mapping function to initialize individuals' positions and employs the idea of genetic evolution into the learner phase. To verify the effectiveness of ETLBO, 20 IEEE congress on Evolutionary Computation benchmark test functions are applied to test its convergence speed and convergence accuracy. Experimental results reveal that ETLBO shows the best convergence accuracy on most functions compared to other state-of-the-art optimization algorithms. In addition, the ETLBO is used to reduce boilers' NOx emissions by optimizing combustion parameters, such as coal supply amount and the air valve. Result shows that ETLBO is well-suited to solve the boiler combustion optimization problem.

Optimal Design of Convolutional Neural Network Architectures Using Teaching–Learning-Based Optimization for Image Classification

Convolutional neural networks (CNNs) have exhibited significant performance gains over conventional machine learning techniques in solving various real-life problems in computational intelligence fields, such as image classification. However, most existing CNN architectures were handcrafted from scratch and required significant amounts of problem domain knowledge from designers. A novel deep learning method abbreviated as TLBOCNN is proposed in this paper by leveraging the excellent global search ability of teaching–learning-based optimization (TLBO) to obtain an optimal design of network architecture for a CNN based on the given dataset with symmetrical distribution of each class of data samples. A variable-length encoding scheme is first introduced in TLBOCNN to represent each learner as a potential CNN architecture with different layer parameters. During the teacher phase, a new mainstream architecture computation scheme is designed to compute the mean parameter values of CNN architectures by considering the information encoded into the existing population members with variable lengths. The new mechanisms of determining the differences between two learners with variable lengths and updating their positions are also devised in both the teacher and learner phases to obtain new learners. Extensive simulation studies report that the proposed TLBOCNN achieves symmetrical performance in classifying the majority of MNIST-variant datasets, displays the highest accuracy, and produces CNN models with the lowest complexity levels compared to other state-of-the-art methods due to its promising search ability.

An improved teaching–learning-based optimization algorithm with a modified learner phase and a new mutation-restarting phase

Teaching–learning-based optimization (TLBO) is a powerful metaheuristic algorithm for solving complex optimization problems pertaining to the global optimum. Many TLBO variants have been presented to improve the local optima avoidance capability and to increase the convergence speed. In this study, a modified learner phase and a new gradient-based mutation strategy are proposed in an improved TLBO algorithm (TLBO-LM). A restarting strategy is adopted to help TLBO-LM escape from local optima and is combined with the gradient-based mutation strategy to build a mutation-restarting phase in an improved TLBO algorithm (LMRTLBO). The modified learner phase integrates a mutation operator to ensure the exploration capability and a dynamic boundary constraint strategy to increase the convergence speed. A greedy gradient information estimation method is developed to accelerate the convergence rate and then is combined with a mutation operator to establish the gradient-based mutation strategy with a good balance between exploration and exploitation. The proposed algorithms are examined by the CEC 2014 and 2020 benchmark functions. The results suggest that both the modified learner phase and gradient-based mutation strategy significantly enhance the exploration and exploitation capabilities of TLBO and that the restarting strategy effectively avoids local optima but sacrifices the exploitation capability to some extent. TLBO-LM and LMRTLBO obtain better results compared with ten algorithms, including TLBO variants, and are competitive compared with five CEC winners.

Modified teaching-learning-based optimization and applications in multi-response machining processes

Many real-world engineering problems such as machining processes are multi-objective optimization problems (MOPs) because multiple performance characteristics are considered to satisfy their contradictory goals. An improved multi-objective teaching–learning-based optimization with refined knowledge sharing mechanisms (IMTLBO-RKSM) is proposed to tackle these MOPs effectively. Pareto dominance concept is first incorporated into IMTLBO-RKSM to handle the tradeoffs of multiple contradictory objectives. Appropriate modifications are incorporated into both teacher and learner phases of IMTLBO-RKSM to emulate to emulate the knowledge sharing processes of classroom more accurately, hence achieving better balancing of exploration and exploitation searches. Particularly, both concepts of Euclidean-distance based teacher assignment scheme and social learning are incorporated into the IMTLBO-RKSM’s teacher phase to derive the unique directional information that can provide better guidance for each learner. The learner phase of IMTLBO-RKSM is also modified by designing two new learning mechanisms known as independent learning and adaptive peer learning, aiming to facilitate different preferences of learners in acquiring new knowledge. The performance of IMTLBO-RKSM is evaluated and compared with six multi-objective optimization methods by using five case studies of multi-response machining problems and twelve MOP benchmark functions. Extensive simulation studies show that IMTLBO-RKSM have more competitive performance than other methods by generating Pareto fronts with better quality in terms of accuracy and diversity of solution members for most tested problems.

INNA: An improved neural network algorithm for solving reliability optimization problems.

The main objective of this paper is to present an improved neural network algorithm (INNA) for solving the reliability-redundancy allocation problem (RRAP) with nonlinear resource constraints. In this RRAP, both the component reliability and the redundancy allocation are to be considered simultaneously. Neural network algorithm (NNA) is one of the newest and efficient swarm optimization algorithms having a strong global search ability that is very adequate in solving different kinds of complex optimization problems. Despite its efficiency, NNA experiences poor exploitation, which causes slow convergence and also restricts its practical application of solving optimization problems. Considering this deficiency and to obtain a better balance between exploration and exploitation, searching procedure for NNA is reconstructed by implementing a new logarithmic spiral search operator and the searching strategy of the learner phase of teaching–learning-based optimization (TLBO) and an improved NNA has been developed in this paper. To demonstrate the performance of INNA, it is evaluated against seven well-known reliability optimization problems and finally compared with other existing meta-heuristics algorithms. Additionally, the INNA results are statistically investigated with the Wilcoxon sign-rank test and Multiple comparison test to show the significance of the results. Experimental results reveal that the proposed algorithm is highly competitive and performs better than previously developed algorithms in the literature.

Q-Learning-Based Teaching-Learning Optimization for Distributed Two-Stage Hybrid Flow Shop Scheduling with Fuzzy Processing Time

Two-stage hybrid flow shop scheduling has been extensively considered in single-factory settings. However, the distributed two-stage hybrid flow shop scheduling problem (DTHFSP) with fuzzy processing time is seldom investigated in multiple factories. Furthermore, the integration of reinforcement learning and metaheuristic is seldom applied to solve DTHFSP. In the current study, DTHFSP with fuzzy processing time was investigated, and a novel Q-learning-based teaching-learning based optimization (QTLBO) was constructed to minimize makespan. Several teachers were recruited for this study. The teacher phase, learner phase, teacher's self-learning phase, and learner's self-learning phase were designed. The Q-learning algorithm was implemented by 9 states, 4 actions defined as combinations of the above phases, a reward, and an adaptive action selection, which were applied to dynamically adjust the algorithm structure. A number of experiments were conducted. The computational results demonstrate that the new strategies of QTLBO are effective; furthermore, it presents promising results on the considered DTHFSP.