Data-driven decision model based on dynamical classifier selection

Non-parametric classification procedures based on a certainty measure and nearest neighbour rule for motor unit potential classification (MUP) during electromyographic (EMG) signal decomposition were explored. A diversity-based classifier fusion approach is developed and evaluated to achieve improved classification performance. The developed system allows the construction of a set of non-parametric base classifiers and then automatically chooses, from the pool of base classifiers, subsets of classifiers to form candidate classifier ensembles. The system selects the classifier ensemble members by exploiting a diversity measure for selecting classifier teams. The kappa statistic is used as the diversity measure to estimate the level of agreement between base classifier outputs, i.e., to measure the degree of decision similarity between base classifiers. The pool of base classifiers consists of two kinds of classifiers: adaptive certainty-based classifiers (ACCs) and adaptive fuzzy k-NN classifiers (AFNNCs) and both utilize different types of features. Once the patterns are assigned to their classes, by the classifier fusion system, firing pattern consistency statistics for each class are calculated to detect classification errors in an adaptive fashion. Performance of the developed system was evaluated using real and simulated EMG signals and was compared with the performance of the constituent base classifiers and the performance of the fixed ensemble containing the full set of base classifiers. Across the EMG signal data sets used, the diversity-based classifier fusion approach had better average classification performance overall, especially in terms of reducing classification errors.

Workshop report

Vision-language integration using constrained local semantic features

A decision model for selecting a suitable supplier is a key to reducing the environmental impact in green supply chain management for high-tech companies. Traditional fuzzy weight average (FWA) adopts linguistic variable to determine weight by experts. However, the weights of FWA have not considered the public voice, meaning the viewpoints of consumers in green supply chain management. This paper focuses on developing a novel decision model for green supplier selection in the One Belt and One Road (OBOR) initiative through a fuzzy weighted average approach with social media. The proposed decision model uses the membership grade of the criteria and sub-criteria and its relative weights, which consider the volume of social media, to establish an analysis matrix of green supplier selection. Then, the proposed fuzzy weighted average approach is considered as an aggregating tool to calculate a synthetic score for each green supplier in the Belt and Road initiative. The final score of the green supplier is ordered by a non-fuzzy performance value ranking method to help the consumer make a decision. A case of green supplier selection in the light-emitting diode (LED) industry is used to demonstrate the proposed decision model. The findings demonstrate (1) the consumer’s main concerns are the “Quality” and “Green products” in LED industry, hence, the ranking of suitable supplier in FWA with social media information model obtained the difference result with tradition FWA; (2) OBOR in the LED industry is not fervently discussed in searches of Google and Twitter; and (3) the FWA with social media information could objectively analyze the green supplier selection because the novel model considers the viewpoints of the consumer.

PurposeThis paper proposes a model based on minority game (MG) theory to support the decision-making regarding the efficient allocation and exploitation of resources/services among the partners of a cloud manufacturing (CMfg) system. CMfg system is a new manufacturing paradigm to share manufacturing capabilities and resources on a cloud platform. The use of a decision model to organize and manage the resources and services provided by the autonomous participants of a CMfg has crucial relevance for the system's effectiveness and efficiency.Design/methodology/approachThis research proposes a noncooperation model based on MG theory. The MG is designed to make decisions on the use of resources/services among the partners of CMfg with private information. A simulation environment was developed to test the efficiency of the proposed decision model. Moreover, an ideal decision model with complete information among the partners was used as a benchmark model.FindingsThe simulation results show how the application of the proposed MG model outperforms the MG model usually proposed in the literature. In particular, the proposed decision model based on private information has an efficiency closer to the ideal model with complete information among the partners of a CMfg.Originality/valueThis paper advances knowledge about the application of MG in the field of CMfg system. The proposed decision-making model based on MG is a promising approach to help enterprises, and especially small and medium enterprises, to participate in CMfg initiatives and to develop their business.

This paper presents a multi-criteria decision model and multi-criteria genetic algorithm (MCGA) approach to improve backbone topology by leveraging the virtual link (VL) system in an hierarchical link-state (LS) routing domain. Given that the sound backbone topology structure has a great impact on the overall routing performance in an hierarchical LS domain, the importance of this research is evident. The proposed decision model is to find an optimal configuration of VLs that properly meets two-pronged engineering goals in installing and maintaining VLs: i.e., operational costs and network reliability. The experiment results clearly indicates that it is essential to the effective operations of hierarchical LS routing domain to consider not only engineering aspects but also specific benefits from systematical layout of VLs, thereby presenting the validity of the decision model and MCGA

This paper proposes an energy-efficient dynamic decision model for wireless multi-sensor network, which is based on the dynamic analysis of the energy consumption characteristics of wireless multi-sensor nodes. We analyze the behaviors of the nodes in wireless multi-sensor network and introduce the existing energy-efficient decision methods, then propose a simple dynamic decision model and prove it theoretically. This paper uses MATLAB 2015 to carry out simulation experiments under the condition of fixed routing protocol based on tree topology and two low-power routing protocols based on mesh topology, and simulation results show that the network lifetime is obviously prolonged. Extending the application of the proposed decision model to aquaculture environmental monitoring system, testing results show that the proposed decision model can effectively reduce the network energy consumption, and be promising when generalized to other applications.

EROS: Ensemble rough subspaces

In this paper, a probabilistic-linguistic-information-driven decision-making method is proposed to help decision makers make decisions with historical data. Historical individual assessments and collective assessments represented by probabilistic linguistic term sets (PLTSs) are used to extract criterion weights to train a decision model in the proposed method. To support the extraction of criterion weights, a new difference measure between PLTSs is developed, which considers semantics of linguistic terms and all possible difference cases except for intersection between PLTSs. The arithmetic, geometric, and quadratic mean operators are developed to iteratively combine individual assessments to generate the aggregated assessments, in which the problem of combinatorial explosion is effectively avoided. The three operators accomplish the training of the decision model based on the extracted criterion weights. The proposed method is used to help radiologists diagnose thyroid nodules, in which its applicability is verified and the performance of the three aggregation operators is compared.

Data-driven collapse strength modelling for the screen pipes with internal corrosion defect based on finite element analysis and tree-based machine learning

Analysis of park-and-ride decision behavior based on Decision Field Theory

In developing countries, conflict between the objectives of environmental policies and those focused on economic development, a low availability of human resources and infrastructure, and the lack of continuity and incentives, contribute to the inability to successfully implement environmental policies. Moreover, in these countries, population growth in the peri-urban areas has resulted in serious water pressures, poor water management, and severe non-point source pollution. The aim of this paper is to implement a procedure of aquifer pollution risk assessment for identifying priority areas in an urban-rural interface based on multi-criteria decision tools. In this sense, a multi-criteria decision model was designed, in terms of environmental and socio-economic criteria and sub-criteria relevant to water resource management, by using the software Criterium Decision Plus 4.0. This model was applied to the northwestern peri-urban area of Mar del Plata City, Argentina, which is embedded in the second most important fruit-horticultural belt of the country. Here, groundwater is the only source of water supply. The proposed decision model allowed establishing a ranking of priority areas for defining action guidelines in order to minimize the risk of pollution of the aquifer. Aquifer pollution hazard criterion mainly contributes to alternatives classified as very high priority whilst in the case of low and very low priority classes, social vulnerability criterion mostly influences the final results. The first alternatives will require an intervention in a short-term time horizon, and the last ones must be monitored in order to avoid their shift to a worse environmental condition.

Hospitals are the first point of contact for people in the face of disasters that interfere with the daily functioning of life and endanger health and social life. All preparations should be made considering the worst possible conditions and the provided service should continue without interruption. In this study, a multi-criteria decision-making model was developed to evaluate disaster preparedness of hospitals. This decision model includes Bayesian best–worst method (BBWM), the VlseKriterijumska Optimizacija I Kompromisno Resenje (VIKOR) and technique for order preference by similarity to ideal solution (TOPSIS) methods. With the proposed decision model, six main criteria and 34 sub-criteria related to disaster preparedness of hospitals were considered. The criteria and sub-criteria evaluated in pairwise comparison manner by the experts were weighted with BBWM. These weight values and the data obtained from the six Turkish hospitals were combined to provide inputs for VIKOR and TOPSIS. In addition, a comparative study and sensitivity analysis were carried out using weight vectors obtained by different tools. BBWM application results show that the “Personnel” criterion was determined as the most important criterion with an importance value of 26%. This criterion is followed by “Equipment” with 25%, “Transportation” with 14%, “Hospital building” and “Communication” with 12%, and “Flexibility” with 11%. Hospital-2 was determined as the most prepared hospital for disasters as a result of VIKOR application. The VIKOR Q value of this hospital was obtained as 0.000. According to the results of the comparative study, Hospital-2 was determined as the most disaster-ready hospital in all six different scenarios. This hospital is followed by Hospital-4 (Q = 0.5661) and Hospital-5 (Q = 0.7464). The remaining rankings were Hospital-6, Hospital-3 and Hospital-1. The solidity of the results was checked with TOPSIS. Based on TOPSIS application results, Hospital-2 was again found the most-ready hospital. The usage of BBWM in this study enabled the expert group’s views to be combined without loss of information and to determine the criteria and sub-criteria weights with less pairwise comparisons in a probabilistic perspective. Via the “Credal ranking”, which is the contribution of BBWM to the literature, the interpretation of the hierarchy between each criterion has been performed more precisely.

Occupant-centric control (OCC) of building systems, which improves energy performance and thermal comfort by considering occupants’ presence, preferences, and activity, has drawn substantial research interest over the past decade. However, most existing studies have focused on evaluating overall performance during a relatively long period, overlooking control robustness during transients.Corner cases in autonomous driving refer to a bug or problem that occurs under unseen, unusual, and perhaps dangerous circumstances. For OCC in buildings, corner cases can be triggered by extreme values or rare patterns of disturbances that are not covered by historical data used to design the controllers. The backbone of many OCC applications is the predictive model, and the root of potential corner cases is that data-driven models sometimes cannot extrapolate well to support decision-making. As the predictive model cannot accurately comprehend the situation or predict what will happen, corner cases could lead to suboptimal or even wrong control actions.Unlike other data-rich domains, building data typically only covers limited operating conditions and therefore is sparse in the feature space. For example, a model for building control trained with data associated with a normal cooling setpoint of 24°C may be unreliable for such higher setpoints as 27 °C. Considering the difficulty in extensive data collection, methods such as physics-informed neural networks have emerged to enhance the extrapolation capability of data-driven models by incorporating physics-based prior knowledge or exerting physical constraints. Yet, despite the claimed merits such as more data-efficient learning and physics-consistent prediction over long horizons, whether and how physics-informed models can help handle the corner cases and improve control robustness remain unknown.To address this knowledge gap, we designed a simulation framework that applies OCC in a multi-zone office building. Corner cases were identified using historical data from a real-world campus building. A group of data-driven predictive models will be developed with different levels of physics integration and tested in OCC simulations. Combining actual and synthetic data, the peer models will be comprehensively examined concerning their predictive capability and control performance. Special attention will be paid to the decision-making process in the corner cases. The experimental results will inform the required level of physics in predictive models to ensure control robustness, leading to two contributions: future development of physics-informed models can be guided in an application-oriented manner, and the actual applications of occupant-centric control can be promoted.

Data-driven analytical models are important tools employed in the petroleum industry to forecast production rates and reserves of petroleum assets. Studies have shown that existing rate-forecasting models project future performance trends by averaging the observed production history data with little or no preferential consideration of the influence of the trends of most recent historical production data. It is also understood that existing empirical data-driven rate forecast models do not account for the uncertainties involved in such future predictions.In this work, new statistically-based data-driven models for production performance forecasting are developed. These models forecast future production performance trends using statistical-exponential smoothing of the historically-observed data, attributing more weights to the most recent historically-observed performance trends. Both linear-exponential and double-exponential smoothing algorithms were considered in the study. The uncertainty analysis of these predictions is evaluated using the history-fit errors derived from the observed data and fits to the history. The models’ accuracy depends on the segment of the observed historical data selected for the initial data fitting and the trends of most recent historical production data—the tail end of the observed data. The application of the proposed statistically-based data driven models was demonstrated on several data sets. The results obtained from the proposed data-driven models compare favourably with the base forecasts derived from existing models such as the Arps’ Decline Curve Analysis techniques. The major advantage of the proposed models over existing models is that predicted forecasts include a range of possible performance trends honouring the observed production history with the associated probabilities and confidence intervals. The major contribution of this work is that coherent statistically-based models have been developed for forecasting production rates and reserves of petroleum assets with uncertainty analysis.