The convergence analysis and specification of the Population-Based Incremental Learning algorithm

The treatment of left turns at signalized intersections drives the development of signal phasing and timing plans and also plays an important role in overall traffic network operations. Accommodating left turns allows for the most direct routeing but reduces intersection capacity, whereas restricting left turns improves capacity but requires some vehicles to travel longer distances. This paper proposes a population-based incremental learning (PBIL) algorithm to determine at which intersections left-turn restrictions should be enacted to maximize a network's operational performance. The performance of each configuration is tested in a micro-simulation environment on both perfect and imperfect square grid networks. Comparison with a partial enumeration of feasible options reveals that the PBIL algorithm is effective at identifying a near-optimal configuration of left-turn restrictions. The resulting configurations suggest that left turns should be generally restricted at intersections that carry the most flow. These intersections typically occur in the central portion of the network when demands are relatively uniform. Doing so helps to provide additional intersection capacity at the locations where it is most necessary, while minimizing the additional travel distance that is incurred due to detours caused by the left-turn restrictions. These provide insight as to how urban traffic networks might be managed to improve network efficiency by only enacting left-turn restrictions at a subset of locations.

The population-based incremental learning (PBIL) is a novel evolutionary algorithm combined the mechanisms of the Genetic Algorithm with competitive learning. In this paper, the influence of the number of selected best solutions on the convergence speed of the PBIL is studied by experiment. Based on experimental results, a PBIL algorithm with elitist strategy, named Double Learning PBIL (DLPBIL), is proposed. The new algorithm learns both the selected best solutions in current population and the optimal solution found so far in the algorithm at same time. Experimental results show that the DLPBIL out-performs the standard PBIL. Both the convergence speed and the solution quality are improved.

A hybrid algorithm (HA) based on population based incremental learning (PBIL) algorithm and zooming algorithm (ZA) is proposed, and its convergence is proved in this paper. In the hybrid algorithm, PBIL algorithm is employed for the evolutionary process as it can accelerate the convergence speed by a reduced time complexity, zooming algorithm is used to improve the PBIL algorithm as it can reduce search space on a large scale, and develop the convergence speed and the precision of solution obviously. The convergent analysis shows that if the population is big, and the parameters are proper, the hybrid algorithm converges to the global optimal solution.

Traffic supervision in telephone networks is a task which needs to determine streams responsible for call losses in a network by comparing their traffic values to nominal values. However, stream traffic values are not measured by the on-line data acquisition system and, hence, have to be computed. We perform this computation by inverting an approximate knowledge based model of stream propagation in circuit-switched networks. This inversion is computed thanks to three evolutionary computation techniques (multiple restart hill-climbing, population-based incremental learning and genetic algorithms) for which both a binary version and a real variant have been experimented with several fitness measures. The final results first point out how the fitness measure choice can impact on their quality. They also show that, in this case, real variants of the algorithms give significantly better results than binary ones.

Performance evaluation of power system stabilizers based on Population-Based Incremental Learning (PBIL) algorithm

In this paper we study two different meta-heuristics to solve a real-word frequency assignment problem (FAP) in GSM networks. We have used a precise mathematical formulation in which the frequency plans are evaluated using accurate interference information coming from a real GSM network. We have developed an improved version of the scatter search (SS) algorithm in order to solve this problem. After accurately tuning this algorithm, it has been compared with a version fixed for the FAP problem of the population-based incremental learning (PBIL) algorithm. The results show that SS obtains better frequency plannings than PBIL for all the experiments performed.

Recently, genetic algorithms (GAs) have attracted great interest owing to efficiency and flexibility against complex robot path planning problems. To accelerate the convergence speed, preceding researches adapted conventional GAs by using problem-specific techniques. However, such approaches increase computational burden and algorithmic complexity, resulting in subsequent additional problems. In this paper, we used Population-Based Incremental Learning (PBIL) algorithm for robot path planning as a probabilistic evolutionary approach. In addition to PBIL, we also proposed the probabilistic model of nodes and the edge bank to generate promising paths. The experimental results demonstrate that the proposed method gave markedly better performance than its conventional counter-parts(GA,kGA,fGA) in terms of success rates and the quality of obtained paths.

Development without any defect is unsubstantial. Timely detection of software defects favors the proper resource utilization saving time, effort and money. With the increasing size and complexity of software, demand for accurate and efficient prediction models is increasing. Recently, search-based techniques (SBTs) have fascinated many researchers for Software Defect Prediction (SDP). The goal of this study is to conduct an empirical evaluation to assess the applicability of SBTs for predicting software defects in object-oriented (OO) softwares. In this study, 16 SBTs are exploited to build defect prediction models for 13 OO software projects. Stable performance measures — GMean, Balance and Receiver Operating Characteristic-Area Under Curve (ROC-AUC) are employed to probe into the predictive capability of developed models, taking into consideration the imbalanced nature of software datasets. Proper measures are taken to handle the stochastic behavior of SBTs. The significance of results is statistically validated using the Friedman test complied with Wilcoxon post hoc analysis. The results confirm that software defects can be detected in the early phases of software development with help of SBTs. This paper identifies the effective subset of SBTs that will aid software practitioners to timely detect the probable software defects, therefore, saving resources and bringing up good quality softwares. Eight SBTs — sUpervised Classification System (UCS), Bioinformatics-oriented hierarchical evolutionary learning (BIOHEL), CHC, Genetic Algorithm-based Classifier System with Adaptive Discretization Intervals (GA_ADI), Genetic Algorithm-based Classifier System with Intervalar Rule (GA_INT), Memetic Pittsburgh Learning Classifier System (MPLCS), Population-Based Incremental Learning (PBIL) and Steady-State Genetic Algorithm for Instance Selection (SGA) are found to be statistically good defect predictors.

The increasing demand of real-time multimedia services makes of quality of service based routing a serious challenge for next-generation networks. The complexity of this NP-complete problem significantly increases with the size of the network. A new evolutionazybased multicast routing algorithm is presented in this paper. It is based on computational intelligence techniques that integrate in an efficient manner the merits of genetic algorithms and the concepts of competitive leazning in the area of artificial neural networks. Population-based incremental learning algorithm is utilized, among other techniques, to construct a delay bounded multicast tree. The proposed algorithm is capable of simultaneously satisfying several key quality of service requirements that aze necessazy for real-time multimedia applications. The main objective of the algorithm is to construct a multicast tree that is characterized by a minimum cost and a bounded end-to-end delay and residual bandwidth. It is shown through ...

In this paper, we introduce a dynamic variant of the Location-Allocation problem: Dynamic Location-Allocation Problem (DULAP). DULAP involves the location of facilities to service a set of customer demands over a defined horizon. To evaluate a solution to DULAP, we propose two holistic metric approaches: Static and Dynamic Approach. In the static approach, a solution is evaluated with the assumption that customer locations and demand remain constant over a defined horizon. In the dynamic approach, the assumption is made that customer demand, and demographic pattern may change over the defined horizon. We introduce a stochastic model to simulate customer population and distribution over time. We use a Genetic Algorithm and Population-Based Incremental Learning algorithm used in previous work to find robust and satisfactory solutions to DULAP. Results show the dynamic approach of evaluating a solution finds good and robust solutions.

Reducing Basel III Capital Requirements with Dynamic Conditional Correlation and Monte Carlo Simulation

The design methodology and the expected performance results of the UHF band feed for the MeerKAT interferometer, which is a precursor to the Square Kilometer Array(SKA), are presented. The feed consists of a wide flare angle axially corrugated conical horn, a horn matching section, a compact orthogonal mode transducer (OMT) and an integrated directional coupler. The MeerKAT dish is a classical Gregorian offset dual-reflector antenna system. A hybrid optimization algorithm comprising of a global search algorithm, namely the Population-Based Incremental Learning algorithm, and a downhill method is used as optimizer for the horn design. The OMT and coupler were optimized with a downhill method. The predicted results show stable radiation patterns over the frequency band with an antenna efficiency above 71 % and a spillover temperature below 20 K at the lowest frequency. The efficiency increases to 78 % at the higher frequencies with a spillover temperature below 12 K. The feed's reflection coefficient is -20 dB and the feed's noise contribution to the system temperature, with the main contributors the OMT and the directional coupler, is below 1.5 K across the frequency band.

Population-based incremental learning (PBIL) has been recently applied to a range of optimization problems in controller designs with promising results. It combines aspects of genetic algorithm with competitive learning. The learning rate in the standard PBIL is generally fixed which makes it difficult for the algorithm to explore the search space effectively. In this paper, the standard PBIL is improved by using a combination of adaptive and fixed learning rate that varies according to the generation. The adaptive-fixed (AF) algorithm can adjust the learning rate automatically according to the degree of evolution of the search. The objective of the power system stabilizer (PSS) design is to achieve adequate stability over a wide range of power system operating conditions. The proposed controller is compared with conventional PBIL with fixed learning rate (PBIL) and tested under various operating conditions. Simulation results show that the AFPBIL based PSS provides a more efficient search capability and gives a better damping and adequate dynamic performance of the system than the conventional PBIL based PSS.

Semi-supervised learning is especially important in data mining applications because it can make use of plentiful unlabeled data to train the high-quality learning models. Semi-Supervised Support Vector Machine (S3VM) is a powerful semi-supervised learning model. However, the high computational cost and non-convexity severely impede the S3VM method in large-scale applications. Although several learning algorithms were proposed for S3VM, scaling up S3VM is still an open problem. To address this challenging problem, in this paper, we propose a new incremental learning algorithm to scale up S3VM (IL-S3VM) based on the path following technique in the framework of Difference of Convex (DC) programming. The traditional DC programming based algorithms need multiple outer loops and are not suitable for incremental learning, and traditional path following algorithms are limited to convex problems. Our new IL-S3VM algorithm based on the path-following technique can directly update the solution of S3VM to converge to a local minimum within one outer loop so that the efficient incremental learning can be achieved. More importantly, we provide the finite convergence analysis for our new algorithm. To the best of our knowledge, our new IL-S3VM algorithm is the first efficient path following algorithm for a non-convex problem (i.e., S3VM) with local minimum convergence guarantee. Experimental results on a variety of benchmark datasets not only confirm the finite convergence of IL-S3VM, but also show a huge reduction of computational time compared with existing batch and incremental learning algorithms, while retaining the similar generalization performance.

This paper proposes a self-adaptive hybrid population-based incremental learning algorithm (SHPBIL) for the m-machine reentrant permutation flow-shop scheduling problem (MRPFSSP) with makespan criterion. At the initial phase of SHPBIL, the information entropy (IE) of the initial population and an Interchange-based search are utilized to guarantee a good distribution of the initial population in the solution space, and a training strategy is designed to help the probability matrix to accumulate information from the initial population. In SHPBIL's global exploration, the IE of the probability matrix at each generation is used to evaluate the evolutionary degree, and then the learning rate is adaptively adjusted according to the current value of IE, which is helpful in guiding the search to more promising regions. Moreover, a mutation mechanism for the probability model is developed to drive the search to quite different regions. In addition, to enhance the local exploitation ability of SHPBIL, a local search based on critical path is presented to execute the search in some narrow and promising search regions. Simulation experiments and comparisons demonstrate the effectiveness of the proposed SHPBIL.