Yu-Wei Wu

Optimizing parameters of support vector machine using fast messy genetic algorithm for dispute classification

Hybrid system is a potential tool to deal with construction engineering and management problems. This study proposes an optimized hybrid artificial intelligence model to integrate a fast messy genetic algorithm (fmGA) with a support vector machine (SVM). The fmGA-based SVM (GASVM) is used for early prediction of dispute propensity in the initial phase of public-private partnership projects. Particularly, the SVM mainly provides learning and curve fitting while the fmGA optimizes SVM parameters. Measures in term of accuracy, precision, sensitivity, specificity, and area under the curve and synthesis index are used for performance evaluation of proposed hybrid intelligence classification model. Experimental comparisons indicate that GASVM achieves better cross-fold prediction accuracy compared to other baseline models (i.e., CART, CHAID, QUEST, and C5.0) and previous works. The forecasting results provide the proactive-warning and decision-support information needed to manage potential disputes.

Fabrication and characterization of thermal conductivity detectors (TCDs) of different flow channel and heater designs

Different flow channel designs and heaters made from different materials were tested for improving the performances of silicon-based thermal conductivity detectors. One of the designs involved an electric heater sandwiched between two identical flow channels for high heat transfer rates. The heater of the other design was suspended over a slot to reduce heat losses. The flow channels were etched in silicon wafers and nickel heating elements were deposited on Pyrex glass, polyimide, and silicon nitride membranes. The transient behaviors of the heaters and the wafer temperatures were measured and analyzed for different voltages. The effects of flow channel design and membrane material on the heat transfer characteristics and sensitivities of the detectors were examined. Simple heat transfer models were developed to aid in understanding and diagnosing detector behaviors and performances. The polyimide heater had the best signal conditions. The warm-up times of the TCDs were found to be primarily dependent upon the package dimensions and properties. The double-channel TCD exhibited 20% higher heat transfer rate compared to the single-channel design, but the sensitivities of these two designs differed only slightly.

A novel time-depended evolutionary fuzzy SVM inference model for estimating construction project at completion

Construction projects frequently face cost overruns during the construction phase. Thus, a proactive approach is essential for monitoring project costs and detection of potential problems. In construction management, Estimate at Completion (EAC) is an indicator for assisting project managers in identifying potential problems and developing appropriate responses. This study utilizes weighted Support Vector Machine (wSVM), fuzzy logic, and fast messy Genetic Algorithm (fmGA) to handle distinct characteristics in EAC prediction. The wSVM is employed as a supervised learning technique that can address the features of time series data. The fuzzy logic is aimed to enhance the model capability of approximate reasoning and to deal with uncertainty in EAC prediction. Moreover, fmGA is utilized to optimize models tuning parameters. Simulation results show that the new developed model has achieved a significant improvement in EAC forecasting.

Improving classification accuracy of project dispute resolution using hybrid artificial intelligence and support vector machine models

Highlights? SVMs have a potential improvement space by integrating hybrid intelligence (HI). ? This work compares the performance of various classification models using HI. ? Disputed construction projects were collected for a real case study. ? This work demonstrates the capability of HI to improve dispute resolution prediction. Support vector machines (SVMs) have been applied successfully to construction knowledge domains. However, SVMs, as a baseline model, still have a potential improvement space by integrating hybrid intelligence. This work compares the performance of various classification models using the combination of fuzzy logic, a fast and messy genetic algorithm, and SVMs. A set of public-private partnership projects was collected as a real case study in construction management. The data were split into mutually independent folds for cross validation. Experimental results demonstrate that the proposed hybrid artificial intelligence system has the best and most reliable classification accuracy at 77.04%, a 24.76% improvement compared with that of SVMs in predicting project dispute resolution (PDR) outcomes (i.e., mediation, arbitration, litigation, negotiation, and administrative appeals) when the dispute category and phase in which a dispute occurs are known during project execution. This work demonstrates the improvement capability of hybrid intelligence in classifying PDR predictions related to public infrastructure projects.

A novel hybrid intelligent approach for contractor default status prediction

Abstract In the construction industry, evaluating the financial status of a contractor is a challenging task due to the myriad of the input data as well as the complexity of the working environment. This article presents a novel hybrid intelligent approach named as Evolutionary Least Squares Support Vector Machine Inference Model for Predicting Contractor Default Status (ELSIM-PCDS). The proposed ELSIM-PCDS is established by hybridizing the Synthetic Minority Over-sampling Technique (SMOTE), Least Squares Support Vector Machine (LS-SVM), and Differential Evolution (DE) algorithms. In this new paradigm, the SMOTE is specifically used to deal with the imbalanced classification problem. The LS-SVM acts as a supervised learning technique for learning the classification boundary that separates the default and non-default contractors. Additionally, the DE algorithm automatically searches for the optimal parameters of the classification model. Experimental results have demonstrated that the classification performance of the ELSIM-PCDS is better than that of other benchmark methods. Therefore, the proposed hybrid approach is a promising alternative for predicting contractor default status.

Novel Genetic Algorithm-Based Evolutionary Support Vector Machine for Optimizing High-Performance Concrete Mixture

AbstractAn effective method for optimizing high-performance concrete mixtures can significantly benefit the construction industry. However, traditional proportioning methods are not sufficient because of their expensive costs, limitations of use, and inability to address nonlinear relationships among components and concrete properties. Consequently, this research introduces a novel genetic algorithm (GA)–based evolutionary support vector machine (GA-ESIM), which combines the K-means and chaos genetic algorithm (KCGA) with the evolutionary support vector machine inference model (ESIM). This model benefits from both complex input-output mapping in ESIM and global solutions with faster convergence characteristics in KCGA. In total, 1,030 data points from concrete strength experiments are provided to demonstrate the application of GA-ESIM. According to the results, the newly developed model successfully produces the optimal mixture with minimal prediction errors. Furthermore, a graphical user interface is uti...

Predicting high-tech equipment fabrication cost with a novel evolutionary SVM inference model

Accurately predicting fabricating cost in a timely manner can enhance corporate competitiveness. This study employs the Evolutionary Support Vector Machine Inference Model (ESIM) to predict the cost of manufacturing thin-film transistor liquid-crystal display (TFT-LCD) equipment. The ESIM is a hybrid model integrating a support vector machine (SVM) with a fast messy genetic algorithm (fmGA). The SVM concerns primarily with learning and curve fitting, while the fmGA is focuses on optimization of minimal errors. Recently completed equipment development projects are utilized to assess prediction performance. The ESIM is developed to achieve the fittest C and @c parameters with minimized prediction error when used for cost estimate during conceptual stages. This study describes an actionable knowledge-discovery process using real-world data for high-tech equipment manufacturing industries. Analytical results demonstrate that the ESIM can predict the costs of manufacturing TFT-LCD fabrication equipment with sufficient accuracy.

Predicting Equilibrium Scour Depth at Bridge Piers Using Evolutionary Radial Basis Function Neural Network

AbstractScouring of bridge piers is a major cause of bridge failure worldwide. Thus, designing safe depths for new bridge foundations and assessing/monitoring the safety of existing bridge foundations are critical to reducing the risk of bridge collapse and the subsequent potential losses in terms of life and property. This paper develops and tests the evolutionary radial basis function neural network (ERBFNN) as a model to forecast scour depth at bridge piers. The ERBFNN is an artificial intelligence (AI) inference model that integrates the radial basis function neural network (RBFNN) and the artificial bee colony (ABC). In the ERBFNN, the RBFNN handles the learning and fitting curves and ABC uses optimization to search for the optimal hidden neuron number Nn and width σ of the Gaussian function. The performance of the ERBFNN is compared with four other AI techniques, including the back-propagation neural network (BPNN), genetic programming (GP), M5 regression tree (M5), and support vector machine (SVM)....

Construction Conceptual Cost Estimates Using Support Vector Machine

Conceptual cost estimate plays an essential role in project feasibility study. In practice, it is performed based on estimators experience. However, due to the inaccuracy of cost estimate, budgeting and cost control are planned and executed inefficiently. Support Vector Machines (SVMs), an Artificial Intelligent technique, is used to conduct the construction cost estimate. The algorithms of SVMs solve a convex optimization problem in a relative short time with satisfied accurate solution. Applying SVMs, the construction conceptual cost estimate model is developed for owners and planners to predict the construction cost of a project. The impact factors of cost estimate are identified through literature review and interview with experts. The cost data of 29 construction projects are used as training cases. Based on the training results, the average prediction error is less than 10% and the computation time is less than 5 minutes. The error is satisfied for the conceptual cost estimate of a project during the planning and conceptual design phase. Case studies show SVMs can efficiently and accurately assist planners to predict the construction cost. cost based on their experiences. Nevertheless, building cost is effected by numerous factors. Some of these factors are full of uncertainty such as geological property and decorative class. Due to such complex and uncertain evaluation process, estimators evaluate building cost using a simple linear manner cannot accurately evaluate the costs. As a result, present building cost estimates are rough. Hsieh(2002) employs the Evolutionary Fuzzy Neural Inference Model (EFNIM) to develop an evolutionary construction conceptual cost estimate model. In the model, Genetic Algorithms are primarily used for optimization; Fuzzy Logic for representing uncertainty and approximate reasoning; and Neural Networks for fuzzy input-output mapping. However the computation run time to search optimal solution takes very long. In order to reduce run time, this study using Support Vector Machine (SVM) to estimate construction cost. The remainder of the paper is organized as follows: In section 2, we introduce Neural Networks (NNs) and Evolutionary Fuzzy Neural Inference Model (EFNIM). In section 3, We define the regression problem and present our approach using SVMs. In section 4 this study compares prediction accuracy and required effort of the SVMs with EFNIM and NNs. Finally in section 5, we conclude and discuss avenues for future work.

Cash flow prediction for construction project using a novel adaptive time-dependent least squares support vector machine inference model

AbstractCash flow information is crucial for the decision making process in construction management. Due to the complexity and the dynamic progress of a construction project, forecasting cash flow demand throughout various phases of the project remains a challenging problem. This article presents a novel inference model, named as Adaptive Timedependent Least Squares Support Vector Machine (LS-SVMAT) for cash flow prediction. In the LS-SVMAT, Least Squares Support Vector Machine (LS-SVM) is integrated with an adaptive time function (ATF) to generalize the inputoutput mapping of cash flow. Since cash flow data are time-dependent, data points recorded in different periods can contribute dissimilarly to the training process of the prediction model. Thus, the role of the ATF is to determine the appropriate weight associated with each data point at a specific time period. By doing so, LS-SVMAT can better deal with the dynamic nature of the time series. Furthermore, to identify the optimal parameters for the inf...