Genetic Programming Approach Research Articles

Genetic programming for high-dimensional imbalanced classification with a new fitness function and program reuse mechanism

Genetic programming (GP) has been successfully applied to classification. However, GP may evolve biased classifiers when encountering the problem of class imbalance. These biased classifiers are often not reliable to be applied to some real-world applications. High dimensionality makes it more difficult for classifiers to effectively separate the majority class and the minority class. The use of GP to handle the joint effect of high dimensionality and class imbalance has not been heavily investigated. In this paper, we propose a GP approach to high-dimensional imbalanced classification, with the goals of increasing the classification performance as well as saving training time. To achieve this goal, a new fitness function is developed to solve the problem of class imbalance, and moreover, a strategy is proposed to reuse previous good GP individuals for improving efficiency. The proposed method is examined on ten high-dimensional imbalanced datasets. Experimental results show that, for high-dimensional imbalanced classification, the proposed method generally outperforms other GP methods and traditional classification algorithms using sampling methods to solve the problem of class imbalance.

Ensemble Adaptive Streaming – A New Paradigm to Generate Streaming Algorithms via Specializations

Video streaming is now ubiquitous in the mobile Internet. This motivated intense research in adaptive streaming algorithms to tackle mobile networks' fluctuating conditions. Our investigations revealed that while existing algorithms can perform well in their intended operating environments, their performance can degrade substantially in other environments. This work tackles this challenge by developing a novel Ensemble Adaptive Streaming (EAS) paradigm to mobile video streaming. As opposed to designing a single streaming algorithm for all network conditions, we argue that different network conditions require different algorithms. We introduce the notion of network differentiators to segregate network conditions into different classes where each class has its own adaptation algorithm designed and optimized specifically for it. An EAS mobile streaming client then selects at runtime the matching adaptation algorithm using the same network differentiators on a per session basis for streaming. We show how EAS can be applied to existing machine-learning approaches to improve their performances. Moreover, to fully exploit EAS's potential we developed the first Genetic Programming approach to evolve adaptive streaming algorithms. The resultant EAS-GP algorithms not only outperformed state-of-the-art algorithms substantially, but also exhibited remarkable robustness over time, location, mobile operators, as well as quality-of-experience metrics.

Time Series QoS Forecasting for Web Services Using Multi-Predictor-Based Genetic Programming

Quality of service (QoS) time series forecasting for web services (WSs) has been studied for over a decade. In recent years, this problem has been investigated in its multistep-ahead version (namely, the problem where the prediction horizon is greater than one) for the long-term rental and use of cloud-based WSs. To solve this multistep-ahead QoS time series forecasting problem, previous research has adopted single-predictor-based strategies and conventional time series methods, such as autoregressive integrated moving average (ARIMA) models and exponential smoothing. In this article, however, we propose applying genetic programming (GP) to search for and evolve a set of multiple predictors, in which each predictor is dedicated to forecasting a specific future time point. Our GP-based approach proposes and tests two types of multiple predictors that differ from the consumed predictor inputs that drive each predictor to produce its QoS forecasting results. In the first type, the input of each predictor is a sequence of fixed previous QoS observations; in the second type, each predictor dynamically consumes the most recent QoS values, which could consist of the forecasting results of its previous predictors). Furthermore, we propose two techniques for our multipredictor-based GP approach, namely, elite individual composition (EIC) and hybrid evolution, and apply them to enhance the forecasting accuracy of our approach. Finally, based on a real-world QoS time series dataset, the proposed approach is validated and compared with several conventional methods to demonstrate its superiority in terms of accuracy; in addition, the effectiveness and efficiency of the proposed multiple predictors and the two techniques are verified in the experiment.

Forecasting Electricity Prices: A Machine Learning Approach

The electricity market is a complex, evolutionary, and dynamic environment. Forecasting electricity prices is an important issue for all electricity market participants. In this study, we shed light on how to improve electricity price forecasting accuracy through the use of a machine learning technique—namely, a novel genetic programming approach. Drawing on empirical data from the largest EU energy markets, we propose a forecasting model that considers variables related to weather conditions, oil prices, and CO2 coupons and predicts energy prices 24 h ahead. We show that the proposed model provides more accurate predictions of future electricity prices than existing prediction methods. Our important findings will assist the electricity market participants in forecasting future price movements.

An Ant Colony Optimization approach for symbolic regression using Straight Line Programs. Application to energy consumption modelling

The increase of energy consumption and their direct effects on pollution and global warming have motivated governments to develop new strategies to promote a better usage of energy. One of the most important aspects related to energy efficiency is the need for a suitable model of energy consumption that can be used to make predictions or to aid experts in high level decision making processes. Symbolic regression techniques can be used to discover an energy consumption model that fits these purposes. Traditionally, the problem of symbolic regression has been solved by using genetic programming approaches to find the algebraic expression that best fits the regression problem data, where each expression is encoded as a tree structure. In previous works, we found that a different approach using Straight Line Programs as a representation technique could provide promising results for symbolic regression, although the size of the resulting algebraic expression might be increased when compared to the traditional approach. This work proposes an Ant Colony Optimization algorithm for Straight Line Programs to solve the problem, and makes a study to compare the approach with traditional genetic programming in a real energy consumption modelling problem.

An ensemble genetic programming approach to develop incipient sediment motion models in rectangular channels

Assimilating unique features of genetic programming (GP) and gene expression programming (GEP), this study introduces a hybrid algorithm which results in promising incipient non-cohesive sediment motion models. The new models use the dimensionless input parameters including relative particle size, relative deposited bed thickness, channel friction factor and channel bed slope to estimate particle Froude number in rectangular channels. The models’ accuracy is tested using different error measures and cross-validated through comparison with that of five empirical models available in the relevant literature. The results showed enhanced accuracy of the proposed models in comparison to the existing ones with concordance correlation coefficient of 0.92 and 0.94 for parsimonious and quasi-parsimonious ensemble GP models, respectively. Such superiority is attributed to the integrated use of flow, fluid, sediment and channel characteristics in the modeling of incipient motion. Although the new algorithm is hybrid, the proposed models are explicit and precise, and thus, motivating to be used in practice.

Genetic Programming for Evolving a Front of Interpretable Models for Data Visualization

Data visualization is a key tool in data mining for understanding big datasets. Many visualization methods have been proposed, including the well-regarded state-of-the-art method t-distributed stochastic neighbor embedding. However, the most powerful visualization methods have a significant limitation: the manner in which they create their visualization from the original features of the dataset is completely opaque. Many domains require an understanding of the data in terms of the original features; there is hence a need for powerful visualization methods which use understandable models. In this article, we propose a genetic programming (GP) approach called GP-tSNE for evolving interpretable mappings from the dataset to high-quality visualizations. A multiobjective approach is designed that produces a variety of visualizations in a single run which gives different tradeoffs between visual quality and model complexity. Testing against baseline methods on a variety of datasets shows the clear potential of GP-tSNE to allow deeper insight into data than that provided by existing visualization methods. We further highlight the benefits of a multiobjective approach through an in-depth analysis of a candidate front, which shows how multiple models can be analyzed jointly to give increased insight into the dataset.

Forecasting stock market return with nonlinearity: a genetic programming approach

The issue whether return in the stock market is predictable remains ambiguous. This paper attempts to establish new return forecasting models in order to contribute on addressing this issue. In contrast to existing literatures, we first reveal that the model forecasting accuracy can be improved through better model specification without adding any new variables. Instead of having a unified return forecasting model, we argue that stock markets in different countries shall have different forecasting models. Furthermore, we adopt an evolutionary procedure called Genetic programming (GP), to develop our new models with nonlinearity. Our newly-developed forecasting models are testified to be more accurate than traditional AR-family models. More importantly, the trading strategy we propose based on our forecasting models has been verified to be highly profitable in different types of stock markets in terms of stock index futures trading.

Towards the use of vector based GP to predict physiological time series

Prediction of physiological time series is frequently approached by means of machine learning (ML) algorithms. However, most ML techniques are not able to directly manage time series, thus they do not exploit all the useful information such as patterns, peaks and regularities provided by the time dimension. Besides advanced ML methods such as recurrent neural network that preserve the ordered nature of time series, a recently developed approach of genetic programming, VE-GP, looks promising on the problem in analysis. VE-GP allows time series as terminals in the form of a vector, including new strategies to exploit this representation. In this paper we compare different ML techniques on the real problem of predicting ventilation flow from physiological variables with the aim of highlighting the potential of VE-GP. Experimental results show the advantage of applying this technique in the problem and we ascribe the good performances to the ability of properly catching meaningful information from time series.

Towards the use of genetic programming in the ecological modelling of mosquito population dynamics

Predictive algorithms are powerful tools to support infection surveillance plans based on the monitoring of vector abundance. In this article, we explore the use of genetic programming (GP) to build a predictive model of mosquito abundance based on environmental and climatic variables. We claim, in fact, that the heterogeneity and complexity of this kind of dataset demands algorithms capable of discovering complex relationships among variables. For this reason, we benchmarked GP performance with state of the art machine learning predictive algorithms. In order to provide a real exploitable model of mosquito abundance, we trained GP and the other algorithms on mosquito collections from 2002 to 2005 and we tested the predictive ability in 2006 collections. Results reveal that, among the studied methods, GP has the best performance in terms of accuracy and generalization ability. Moreover, the intrinsic feature selection and readability of the solution provided by GP offer the possibility of a biological interpretation of the model which highlights known or new behaviours responsible for mosquito abundance. GP, therefore, reveals to be a promising tool in the field of ecological modelling, opening the way to the use of a vector based GP approach (VE-GP) which may be more appropriate and beneficial for the problems in analysis.

Learning ensembles of priority rules for online scheduling by hybrid evolutionary algorithms

This paper studies the computation of ensembles of priority rules for the One Machine Scheduling Problem with variable capacity and total tardiness minimization. Concretely, we address the problem of building optimal ensembles of priority rules, starting from a pool of rules evolved by a Genetic Programming approach. Building on earlier work, we propose a number of new algorithms. These include an iterated greedy search method, a local search algorithm and a memetic algorithm. Experimental results show the potential of the proposed approaches.

Evolving Suspiciousness Metrics From Hybrid Data Set for Boosting a Spectrum Based Fault Localization

Spectrum Based Fault Localization (SBFL) uses different metrics called risk evaluation formula to guide and pinpoint faults in debugging process. The accuracy of a specific SBFL method may be limited by the used formulae and program spectra. However, it has been demonstrated recently that Genetic Programming could be used to automatically design formulae directly from the program spectra. Therefore, this article presents Genetic Programming approach for proposing risk evaluation formula with the inclusion of radicals to evolve suspiciousness metric directly from the program spectra. 92 faults from Unix utilities of SIR repository and 357 real faults from Defect4J repository were used. The approach combines these data sets, used 25% of the total faults (113) to evolve the formulae and the remaining 75% (336) to validate the effectiveness of the metrics generated by our approach. The proposed approach then uses Genetic Programming to run 30 evolution to produce different 30 metrics. The GP-generated metrics consistently out-performed all the classic formulae in both single and multiple faults, especially OP2 on average of 2.25% in single faults and 3.42% in multiple faults. The experiment results conclude that the combination of Hybrid data set and radical is a good technique to evolve effective formulae for spectra-based fault localization.

Learning Regular Expressions Using XCS-Based Classifier System

Evolutionary machine learning research aims to develop classifier systems that can solve complex and hard tasks. This paper addresses the problem of inferring a regular expression from a given set of strings for automating the task of information extraction. To the best of our knowledge, this paper is the first to propose the extension of accuracy-based classifier system XCS to learn the regular expressions for text extraction. This new system named as XCSREA includes tree-like code fragments to learn regular expressions. The genetic algorithm in action sets uses two-point crossover with uniform mutation and Roulette wheel parent selection method. Seven different datasets, each with three different lengths, are used to compare the performance of the proposed model with standard genetic programming (GP) approach. The experimental results demonstrate that XCSREA outperforms standard GP approach when sufficiently large numbers of classifiers are used.

Highly Effective Security Techniques in OSN Based on Genetic Programming Approach

Assurance is one of the grinding centers that enhances when trades get mediated in Online Community Techniques (Online Social Networks). Diverse gatherings of utilization innovation specialists have limited the 'OSN security issue' as one of surveillance, institutional or open security. In taking care of these issues they have moreover overseen them just as they were person. We adapt that the elite security issues are caught and that evaluation on genuine feelings of serenity in Online Social Networks would advantage from a more exhaustive method. Nowadays, points of interest systems mean a critical piece of relationship; by losing security, these organizations will decrease a ton of pleasant areas to see as well. The inside motivation behind subtle elements security (Information Security) is risk control. There are a great deal of discovering works and exercises in security danger control (ISRM, for example, NIST 800-30 and ISO/IEC 27005. Regardless, only few works of appraisal focus on Information Security danger diminishment, while the signs depict normal determinations and suggestions. They don't give any use ideas concerning ISRM; truth be told diminishing the Information Security dangers in questionable conditions is cautious. Subsequently, this papers joined an acquired counts (GA) for Information Security danger loss of weaknesses. Finally, the parity of the associated system was broke down through a reflection.

Forecasting of TBM advance rate in hard rock condition based on artificial neural network and genetic programming techniques

The efficiency of tunnel boring machine (TBM) is regarded as a key factor in successfully undertaking any mechanical tunneling project. In fact, an accurate forecasting of TBM performance, especially in a specified rock mass condition, can minimize capital costs and scheduling for tunnel excavation. This study puts an effort to propose two accurate and practical predictive models of TBM performance via artificial neural network (ANN) and genetic programming (GP) approaches. To set a certain prediction target for the proposed models, the advance rate (AR) of TBM is considered as its performance metric. For modeling purpose, a large experimental database containing 1286 data sets was set up as the result of conducting site investigation operations for a tunneling project in Malaysia, called the Pahang–Selangor Raw Water Transfer Tunnel and performing a number of laboratory tests on the collected rock samples. To design the desired intelligent models of AR based on the training and test patterns, a mix of rock and machine characteristics with the most influence on AR has been used as input parameters, i.e., rock quality designation (RQD), uniaxial compressive strength (UCS), rock mass rating (RMR), Brazilian tensile strength (BTS), thrust force (TF), and revolution per minute (RPM). In addition, a series of statistical indices, such as root mean squared error (RMSE), determination coefficient (R-square), and variance account for (VAF) are utilized to evaluate and compare the prediction precision of the developed models. Based on the simulation results and the computed values of indices, it is observed that the proposed GP model with the training and test RMSE values 0.0427 and 0.0388, respectively, performs noticeably better than the proposed ANN model giving RMSE values 0.0509 and 0.0472 for the training and test sets, respectively. Additionally, a parametric analysis has been conducted on the proposed GP model to further verify its generalization capability. The obtained results demonstrate that this GP-based model could provide a new applicable equation for accuratly predicting TBM performance.

Feature Selection using Genetic Programming

Extracting useful and novel information from the large amount of collected data has become a necessity for corporations wishing to maintain a competitive advantage. One of the biggest issues in handling these significantly large datasets is the curse of dimensionality. As the dimension of the data increases, the performance of the data mining algorithms employed to mine the data deteriorates. This deterioration is mainly caused by the large search space created as a result of having irrelevant, noisy and redundant features in the data. Feature selection is one of the various techniques that can be used to remove these unnecessary features. Feature selection consequently reduces the dimension of the data as well as the search space which in turn increases the efficiency and the accuracy of the mining algorithms. In this paper, we investigate the ability of Genetic Programming (GP), an evolutionary algorithm searching strategy capable of automatically finding solutions in complex and large search spaces, to perform feature selection. We implement a basic GP algorithm and perform feature selection on 5 benchmark classification datasets from UCI repository. To test the competitiveness and feasibility of the GP approach, we examine the classification performance of four classifiers namely J48, Naives Bayes, PART, and Random Forests using the GP selected features, all the original features and the features selected by the other commonly used feature selection techniques i.e. principal component analysis, information gain, relief-f and cfs. The experimental results show that not only does GP select a smaller set of features from the original features, classifiers using GP selected features achieve a better classification performance than using all the original features. Furthermore, compared to the other well-known feature selection techniques, GP achieves very competitive results.

A genetic programming-based approach to identify potential inhibitors of serine protease of Mycobacterium tuberculosis.

Aim: We applied genetic programming approaches to understand the impact of descriptors on inhibitory effects of serine protease inhibitors of Mycobacterium tuberculosis (Mtb) and the discovery of new inhibitors as drug candidates. Materials & methods: The experimental dataset of serine protease inhibitors of Mtb descriptors was optimized by genetic algorithm (GA) along with the correlation-based feature selection (CFS) in order to develop predictive models using machine-learning algorithms. The best model was deployed on a library of 918 phytochemical compounds to screen potential serine protease inhibitors ofMtb. Quality and performance of the predictive models were evaluated using various standard statistical parameters.Result:The best random forest model with CFS-GA screened 126 anti-tubercular agents out of 918 phytochemical compounds. Also, genetic programing symbolic classification method is optimized descriptors and developed an equation for mathematical models. Conclusion: The use of CFS-GA with random forest-enhanced classification accuracy and predicted new serine protease inhibitors of Mtb, which can be used for better drug development against tuberculosis.

Strength evaluation of granite block samples with different predictive models

Over the last decade, application of soft computing techniques has rapidly grown up in different scientific fields, especially in rock mechanics. One of these cases relates to indirect assessment of uniaxial compressive strength (UCS) of rock samples with different artificial intelligent-based methods. In fact, the main advantage of such systems is to readily remove some difficulties arising in direct assessment of UCS, such as time-consuming and costly UCS test procedure. This study puts an effort to propose four accurate and practical predictive models of UCS using artificial neural network (ANN), hybrid ANN with imperialism competitive algorithm (ICA–ANN), hybrid ANN with artificial bee colony (ABC–ANN) and genetic programming (GP) approaches. To reach the aim of the current study, an experimental database containing a total of 71 data sets was set up by performing a number of laboratory tests on the rock samples collected from a tunnel site in Malaysia. To construct the desired predictive models of UCS based on training and test patterns, a combination of several rock characteristics with the most influence on UCS has been used as input parameters, i.e. porosity (n), Schmidt hammer rebound number (R), p-wave velocity (Vp) and point load strength index (Is(50)). To evaluate and compare the prediction precision of the developed models, a series of statistical indices, such as root mean squared error (RMSE), determination coefficient (R2) and variance account for (VAF) are utilized. Based on the simulation results and the measured indices, it was observed that the proposed GP model with the training and test RMSE values 0.0726 and 0.0691, respectively, gives better performance as compared to the other proposed models with values of (0.0740 and 0.0885), (0.0785 and 0.0742), and (0.0746 and 0.0771) for ANN, ICA–ANN and ABC–ANN, respectively. Moreover, a parametric analysis is accomplished on the proposed GP model to further verify its generalization capability. Hence, this GP-based model can be considered as a new applicable equation to accurately estimate the uniaxial compressive strength of granite block samples.

On the use of the genetic programming for balanced load distribution in software-defined networks

As a new networking paradigm, Software-Defined Networking (SDN)enables us to cope with the limitations of traditional networks. SDN uses a controller that has a global view of the network and switch devices which act as packet forwarding hardware, known as “OpenFlow switches”. Since load balancing service is essential to distribute workload across servers in data centers, we propose an effective load balancing scheme in SDN, using a genetic programming approach, called Genetic Programming based Load Balancing (GPLB). We formulate the problem to find a path: 1) with the best bottleneck switch which has the lowest capacity within bottleneck switches of each path, 2) with the shortest path, and 3) requiring the less possible operations. For the purpose of choosing the real-time least loaded path, GPLB immediately calculates the integrated load of paths based on the information that receives from the SDN controller. Hence, in this design, the controller sends the load information of each path to the load balancing algorithm periodically and then the load balancing algorithm returns a least loaded path to the controller. In this paper, we use the Mininet emulator and the OpenDaylight controller to evaluate the effectiveness of the GPLB. The simulative study of the GPLB shows that there is a big improvement in performance metrics and the latency and the jitter are minimized. The GPLB also has the maximum throughput in comparison with related works and has performed better in the heavy traffic situation. The results show that our model stands smartly while not increasing further overhead.

Design and simulation of nano-arbiters using quantum-dot cellular automata

Arbiters are the essential components of the Network-On-Chip (NOC) systems and are used to resolve the contention problem where multiple requests must be handled for shared resources. On the other hand, with the ever-increasing downsizing trend in the fabrication technology, Quantum-dot Cellular Automata (QCA) with its nano scales and very low power consumption is a promising candidate for implementing future NOCs. In the current work, we design and simulate nano-arbiters using QCA with the following contributions: i) The 2-bit Basic Round Robin Arbiter (RRA) and the 2-bit Ping Pong Arbiter (PPA) are designed and simulated; ii) A solution for an erroneous condition found in the original circuit of RRA is reported and fixed; iii) We use Cartesian Genetic Programming (CGP) approach to simplify the RRA and PPA designs; iv) In order to leverage our QCA designs, we apply a more realistic clock distribution (2-DW clocking) and report the results. At the end, a one-to-one comparison of the two arbiters designed with QCA will be presented using such benchmarks as area, latency, etc. Our results show that in the 2-bit input mode, the PPA arbiter has the best overall performance.