An improved teaching-learning-based optimization: briefly survey
Ebtehal Akeel Hamed
1,
Inas Kadhim jebur
2, Mustafa Asaad Hasan
3, Rusul A. Salman
4,
Hayder A. Jawdhari
5 1Al Qasim Green University, Iraq 2Al Qasim Green University, Iraq 3University of Thi-Qar4Al Qasim Green University, Iraq 4haider.satar@uoqasim.edu.iq
Article History: Received: 10 January 2021; Revised: 12 February 2021; Accepted: 27 March 2021; Published online: 10 May 2021
Abstract: Teaching-learning-based optimization algorithm (TLBO) considers a newly advanced heuristic algorithm that depends on the physical appearance of the teaching-learning procedure. Its goal at the reduction of facilely come down into local optimum when working on complicated optimization difficulties that have high dimensional, in this paper, we will review a viewpoint many studies that involved an improved or enhanced teaching-learning-based optimization, that interested by improved learner phase which included the population to avert the potential about overthrowing toward a local optimum. Also, strategies based on the Gaussian perturbation ban the TLBO algorithm from falling within local minima. In this paper also presented the role of an Improved Teaching-learning-based Optimization Algorithm to Determine Job Shop Scheduling issues.
1. Introduction
Algorithm of Teaching-Learning-Based Optimization (TLBO) [1], that seems the same traditional approach from teaching procedure of teacher and studying or examine procedure of students, depend on the physical phenomenon of learning, Is a modern suggested heuristic algorithm that is important for optimization applications [2-3]. Has rapid convergence velocity and high searching accuracy, the population size with iterations number it a most important requirement of the TLBO to be set, it always ignoring parameters in its work. Since the TLBO started its work or put forward, it has succeeded in many problems solving since it has taken big attention from scientists [4]. In numerous fields, TLA proved its success, these domains inclusive of mechanical schemes and mechanistic processing optimization [5], heat exchange optimization [6], clustering of data optimization [7]. In [8] suggested (ET LBO) by proposing elite renewal strategy, and it has been obvious to execute fully in fixing complicated not restricted optimization use manner. Some suggested an Elitist Teaching-Learning Opposition-based algorithm (ETLOBA); the algorithm has developed in the search precision and convergence rate disparity to TLBO [9]. Others suggested the neighborhood search as enhanced teaching learning-based optimization algorithm with (NSTLBO) [10], that doing swimmingly to perform exploration during the employment of ANN. To develop algorithm search ability, proposed the social character of PSO on the teacher point of TLBO [11]. To increase the focused on the improvement proposed a developed TLBO through offering special notions about a number of teachers [12], adaptive teaching operators, tutorial exercise, and self-motivated learning. These improved algorithms should obtain outcomes that may classify as good results, but there are yet several troubles. The riser complication of the algorithm and reducing variety of the population are obtained from Improved algorithms could easily, which forces the algorithm to drop within the
local optimal or early convergence. getting fix the mentioned issues, someone search for getting the improved teaching-learning-based optimization (ITLBO) through offering strategy in the second-teaching approach, the improved learner phase, and self-exploration study of teacher, that improves traditional TLBO in regards to the speed regarding the convergence, the precision of the algorithm and the strength jump to outward from local optimal.
Heuristic research methods have been widely used in solving optimization problems in the last two decades. These methods are optimization problems with any process in nature. It is based on establishing a relationship between. Genetic algorithms (GA), ant colony optimization (ACO), particle swarm optimization (PSO), harmony search (HS), artificial bee colony optimization (ABC), learning-teaching based optimization (TLBO) are among the most important heuristic research methods. Genetic algorithms are based on the survival of the individual most suitable for the conditions existing in nature. Apply to optimization problems. Ant colony optimization mimics the strategies ants employ to find the shortest path between their nests and food sources. Particle swarm optimization establishes a similarity between the behavior of habitat colonies such as insect swarm, bird swarm, or fish swarm
and optimization problems. In the harmony search method, it is based on establishing a similarity between the way musicians follow to obtain the best harmony and optimization problems and develop solution methods. Artificial bee colony algorithm; working mechanism in bee colonies and It is based on imitating the strategies they follow to obtain nectar. Teaching-learning-based optimization method is a method that models the effect of a teacher on students and the interactions of students with each other. Harmony search (HS), which is considered a powerful heuristic research method in this survey was explain. Harmony search method was first used [16] has been suggested Method building systems It was also used in optimization and gave very successful results.
2. Teaching-Learning-Based Optimization (TLBO)
The TLBO algorithm relies upon the impact that based effect of the teacher on learners in the class. Such as algorithms nature-inspired, the TLBO is as well a population-based approach which utilizes as a population of explications to forward to the global resolution, however the way doesn't have a user-specified parameter. A collection of students is looked at as the population (M). Each student is looked as a person (M , = 1 : Z, wherever Z is the population volume). Through the TLBO algorithm, various topics presented to students are seen being diverse scheme elements. The outcome of the learning of a student is will be similar to the fitness (function value (M), = 1 : Z) as the same is the case in different optimization algorithms. The instructor is looked at as a better-informed character within the class that contributes to the teacher's information in addition to the students to increase the grades of a class. The goodness of the students is estimated through the average of the learner's grade in a class. The procedure of TLBO is split into a couple of sections. The prime section is composed from the “Teacher phase” and the other section comprise of the “Learner phase”. The “Teacher phase” refers to studying from the tutor and the “Learner phase” refers to learning over the intercommunication amongst students.
Implementations of TLBO
Preparation
Tracking whole symbols utilized to characterized the TLBO
H: the number of students in class in other meaning “class volume” J: the number of seminars proposed to the students
MAXIT: highest admissible repetitions ‘numbers
The population M is at random begin through a search Space surrounded over a matrix of H & J (rows with columns). The jth parameter of the ith learner has defined values in an unarranged approach based on the below equation (1):
Where rand assigns a regularly distributed stochastic variable with domain (0, 1), xminj and xmaxj indicate the lowest and highest rate for jth parameter. The parameters of ith student for the generation g are yielded as:
A. Teacher phase
In this approach, the most suitable explication in the whole population is viewed as the teacher, and the teacher gives the knowledge to the students to grow the average of the result of the course students. Assume xi = (xi1,…..,xid,….xiD) is the location of the ith student, the student including the most suitable fitness is specified being the teacher xteacher, and the average location of a course or class with NP students may describe as Xmean =(1/NP) ∑𝑁𝑃𝑥𝑖
𝑖=1 . The location of every student is modernized through the next equation:
The teacher phase duplicates teachers in the classroom when doing the teaching process. Through this phase, select student of better fitness rates to take the role of teacher for that cycle, the average knowledge level of whole students will rise through a teacher. However, the process of study for each student depending on the capacity of learners and the level of quality of teaching offered over a teacher.
Based on fundamental of variation within the current and the average of whole learners, the up doing or result is modified as stated by the following expression:
Xi
new=Xi + r (Xteacher –TF . mean) (3)
Whither Xi
new are indicate the current location of the ith learner behind teaching by teacher and r is a stochastic value between [0, 1]. The amount of TF is at irregular established each 1 or 2 through the next phrase:
The marks of the whole learners are weighted through fitness measures. To modify the individual following to the teacher phase, on condition that the modern individual Xinew is better than the traditional the Xi, the traditional one is kept.
B. Learner phase
The learner phase imitates the procedure that occurs in investigations and discussions amongst learners. Whole students react at random with different learners to growing their information. The learner phase is expressed as:
Whether r is a stochastic value between [0,1], f(Xi) with f(Xj) are considered the fitness rate of the ith student Xi and the jth student Xj.
The update of an individual is following the student phase whether the current Xinew is better than the traditional, the traditional position Xix is displaced with the modern; oppositely, the traditional one is kept.
3. Improved Teaching-Learning Based Optimization based Gaussian perturbation strategy.
In [13] presented a Gaussian perturbation strategy to stave off TLBO algorithm of falling inside local minima. Furthermore, the opposition-based learning procedure is used in the learning phase to get stretching of the search space. Because the exist bestead solution (that mean better individual in population) can lead the algorithm to look much better solution, to perform this strategy should select a quarter of the best solutions to construct Gaussian distribution, to create a new or modern solution that can be based on the Gaussian distribution to be done. Then have should compare the two solutions (new solution and the current solutions) whether that solution is most suitable than the current best solution, then the new solution exchanges the current best solution. Below can clarify the details:
Level 1: Picked m more qualified learners of the current group of learners.
Level 2: calculate the average of the picked m learners and the standard deviation of the picked m learners depending on the expression:
Whither μj and σ j is the average and standard deviation of the proportions parameter.
Level 3: used the Gaussian distribution with an average of μj in addition to a standard deviation of σ j to create a new students xe1.
Level 4: Matched the fitness of xe1 and pg, pick the better one.
(ITLBO)ALGORITHM SOLVING ECONOMIC LOAD DISPATCH PROBLEMS
In [14] Job workshop scheduling problem (JSP) could be an emphatically NP-hard conjugational optimization issue. Consider troublesome to fix the issue to a particular ideal in a sensible time. Teaching learning-based optimization (TLBO) calculation may be a different populace situated meta-heuristic calculation. It has been shown the TLBO encompasses each significant possibility during matched to the most popular heuristic calculations to the scheduling issues. The common TLBO is promoted to developed diversification in addition to
intensification during exploring solutions for JSP. The supplements include altering the approach of coding, expanding the number of teachers, presenting modern students and implementing local search surrounding probably optimal solutions. Scheduling significant difficulties in the area of making administration and is at most interested in discovering series of duties on assigned machines to reduce the finishing period of implementation of all jobs (i.e., makespan). In JSP, orders that jobs pass over machines are diverse. The hardness is how to discover a better schedule from all feasible schedules [15]. While searching the solution space, the algorithm's calculation period of implementation rises more and more rapidly with the growth of jobs and computers. Therefore, JSP is NP-hard. There are some causes for choosing the TLBO to fix JSP. First, particularly couple of mathematical expressions utilized to generate different solutions in TLBO. Secondary, TLBO consider as free-parameters (ignoring parameters in its work) for that more succinct testing and setting energy for work required. In all of the above [14] offered several improvements to the common TLBO. First, the rise of learner's difference through altering the coding approach. Second, rise number of teachers to expanded variety of convergence. This rising due to learners converges towards their teacher rapidly; third, they generated new learners depending on several principles to enhance the difference of students. Fourth, they presented a local search around potentially optimal solutions.
4. conclusion
In this paper, we review an overview of studies about the TLBO and its improvement, also to increase the improvement the overall achievement of TLBO. Reviewed the works of the teacher phase, the second-teaching approach, and the self-exploration survey of teachers were. And more major learning approaches are explained with the field of learner phase. Furthermore, know some techniques grow the variety of the population.
Also we clarify pieces of information on teaching-learning based optimizations, namely ITLBO algorithms, for decoding unlimited issues of optimization. The position updating operation is incorporated in the introduced four levels. Used the position updating process is to modify the local and global search ability of the original TLBO algorithm.
Also, explained the modern population oriented meta-heuristic algorithm concerning to fix optimization issues. While used it to do JSP, which the studies on this topic have proved that the population convergence scale is quick in addition to simply to trap within local optimum. This overcame these problems, by using improvements and simulations utilizing benchmark issues.
In future search, the self-learning pattern of the teacher and learners should be developed because that contributes to improving the performance of the ITLBO. Over time, the algorithm is intended in the field of multi-objective restraint optimization to improve the execution of ITLBO in the congregation problems.
5. References
1. Rao, R. V., Savsani, V. J., & Vakharia, D. P. (2011). Teaching–learning-based optimization: a novel method for constrained mechanical design optimization problems. Computer-Aided Design, 43(3), 303-315.
2. Rao, R. V., Savsani, V. J., & Vakharia, D. P. (2012). Teaching–learning-based optimization: an optimization method for continuous non-linear large scale problems. Information sciences, 183(1), 1-15.
3. Niknam, T., Azizipanah-Abarghooee, R., & Narimani, M. R. (2012). A new multi objective optimization approach based on TLBO for location of automatic voltage regulators in distribution systems. Engineering Applications of Artificial Intelligence, 25(8), 1577-1588.
4. Chen, X., Mei, C., Xu, B., Yu, K., & Huang, X. (2018). Quadratic interpolation based teaching-learning-based optimization for chemical dynamic system optimization. Knowledge-Based Systems, 145, 250-263. 5. Li, X., Tang, K., Omidvar, M. N., Yang, Z., Qin, K., & China, H. (2013). Benchmark functions for the CEC
2013 special session and competition on large-scale global optimization. gene, 7(33), 8.
6. Rao, R., & Patel, V. (2012). An elitist teaching-learning-based optimization algorithm for solving complex constrained optimization problems. International Journal of Industrial Engineering Computations, 3(4), 535-560.
7. S. Meri Al Absi, A. Hasan Jabbar, S. Oudah Mezan et al., An experimental test of the performance enhancement of a Savonius turbine by modifying the inner surface of a blade, Materials Today: Proceedings, https://doi.org/10.1016/j.matpr.2020.12.309
8. Y Satapathy, S. C., & Naik, A. (2011, December). Data clustering based on teaching-learning-based optimization. In International conference on swarm, evolutionary, and memetic computing (pp. 148-156). Springer, Berlin, Heidelberg.
9. Woo, D. K., Choi, J. H., Ali, M., & Jung, H. K. (2011). A novel multimodal optimization algorithm applied to electromagnetic optimization. IEEE transactions on magnetics, 47(6), 1667-1673.
10. Rajasekhar, A., Rani, R., Ramya, K., & Abraham, A. (2012, October). Elitist teaching learning opposition based algorithm for global optimization. In 2012 IEEE International Conference on Systems, Man, and Cybernetics (SMC) (pp. 1124-1129). IEEE.
11. Wang, L., Zou, F., Hei, X., Yang, D., Chen, D., & Jiang, Q. (2014). An improved teaching–learning-based optimization with neighborhood search for applications of ANN. Neurocomputing, 143, 231-247.
12. Zou, F., Chen, D., & Wang, J. (2016). An improved teaching-learning-based optimization with the social character of PSO for global optimization. Computational intelligence and neuroscience, 2016.
13. Rao, R. V., & Patel, V. (2013). Multi-objective optimization of heat exchangers using a modified teaching-learning-based optimization algorithm. Applied Mathematical Modelling, 37(3), 1147-1162.
14. Zhao, X. H. (2015, July). Improved teaching-learning based optimization for global optimization problems. In 2015 34th Chinese Control Conference (CCC) (pp. 2639-2644). IEEE.
15. Li, L., Weng, W., & Fujimura, S. (2017, May). An improved teaching-learning-based optimization algorithm to solve job shop scheduling problems. In 2017 IEEE/ACIS 16th International Conference on Computer and Information Science (ICIS) (pp. 797-801). IEEE.
16. Anandaraman, C. (2011). An improved sheep flock heredity algorithm for job shop scheduling and flow shop scheduling problems. International Journal of Industrial Engineering Computations, 2(4), 749-764.
