Search

Multi-class classification is an important and challenging research topic with many real-life applications. The problem is much harder than the classical binary classification, especially when the given data set is imbalanced. Hidden nonlinear patterns in the data set can further complicate the task of multi-class classification. In this paper, we propose a kernel-free least squares twin support vector machine for multi-class classification. The proposed model employs a special fourth order polynomial surface, namely the double well potential surface, and adopts the "one-verses-all" classification strategy. An ℓ 2 regularization term is added to accommodate data sets with different levels of nonlinearity. We provide some theoretical analysis of the proposed model. Computational results using artificial data sets and public benchmarks clearly show the superior performance of the proposed model over other well-known multi-class classification methods, in particular for imbalanced data sets. [ABSTRACT FROM AUTHOR]

Revenue management (RM) is essential for a wide range of industries such as airlines, hotels, cruise lines, fashion, and seasonal retail. This paper focuses on the multi-perishable-product dynamic pricing (MPPDP) problem, a significant research field in RM, where a company sells multiple interactive and perishable products over a limited selling window without replenishment. Most studies in this field assume customer behavior, which is modeled by demand function, is known in advance. Even when considering uncertainty in customer behavior, most studies still assume the mathematical form or structural properties of the underlying demand function are known in advance. However, these assumptions are usually inconsistent with the actual market situation. Recently, Reinforcement Learning (RL), a potent technique for handling sequential decision-making problems, has been increasingly applied to solve complex dynamic pricing problems without relying on any assumption about demand functions. However, the curse of dimensionality poses a challenge for currently used centralized RL algorithms when solving the MPPDP problem due to the exponential expansion of the joint price space with the number of products. To address this issue, our paper proposes a distributed dynamic pricing framework and innovatively models the MPPDP problem as a Fully Cooperative Markov Game solved by Multi-Agent Reinforcement Learning (MARL). Additionally, we use counterfactual baselines to design appropriate agent-specific reward signals that facilitate faster learning for the agents in our established multi-agent cooperative system. Finally, two MARL-based distributed dynamic pricing algorithms, Counterfactual Q-learning, and Counterfactual DQN, are proposed for the MPPDP problem. Through the case studies on four computer-simulated markets, we show that our algorithms can alleviate the curse of dimensionality faced by centralized RL algorithms, expedite the learning process, and demonstrate satisfactory performance without relying on any assumption about demand functions. In conclusion, our work provides an effective MARL-based distributed dynamic pricing framework and algorithms for companies to efficiently price their multiple perishable products in modern highly uncertain markets. • Propose a distributed pricing framework for multi-perishable products. • Model multi-product dynamic pricing problem as a Fully Cooperative Markov Game. • Design agent-specific reward functions to tackle the credit assignment problem. • Propose two efficient and effective distributed dynamic pricing algorithms. [ABSTRACT FROM AUTHOR]

We suggest a new greedy strategy for convex optimization in Banach spaces and prove its convergence rates under a suitable behavior of the modulus of uniform smoothness of the objective function. We show that this algorithm is a generalization of the recently discovered Rescaled Pure Greedy Algorithm for approximation in Hilbert spaces. [ABSTRACT FROM AUTHOR]

In binary classification, kernel-free quadratic support vector machines are proposed to avoid difficulties such as finding appropriate kernel functions or tuning their hyper-parameters. Furthermore, Universum data points, which do not belong to any class, can be exploited to embed prior knowledge into the corresponding models to improve the general performance. This paper designs novel kernel-free Universum quadratic surface support vector machine models. Further, this paper proposes the ℓ 1 norm regularized version that is beneficial for detecting potential sparsity patterns in the Hessian of the quadratic surface and reducing to the standard linear models if the data points are (almost) linearly separable. The proposed models are convex, so standard numerical solvers can be utilized to solve them. Moreover, a least squares version of the ℓ 1 norm regularized model is proposed. We also design an effective tailored algorithm that only requires solving one linear system. Several theoretical properties of these models are then reported and proved as well. The numerical results show that the least squares version of the proposed model achieves the highest mean accuracy scores with promising computational efficiency on some artificial and public benchmark data sets. Some statistical tests are conducted to show the competitiveness of the proposed models. [ABSTRACT FROM AUTHOR]

The strip steel has been widely used in the manufacturing industry. Defects on the surface are main factors to determine the quality of strip steel. Due to the various shapes of the defects and background interference, the CNN-based algorithm cannot give full play to its best performance. In this paper, a defect detection module, named detection network with multiscale context (MSC-DNet), is proposed to localize the precise position of defect and classify the specific category of surface defects. In MSC-DNet, a parallel architecture of dilated convolution (PADC) with different dilation rate is built up to capture the multi-scale context information containing multiscale defects. Furthermore, a feature enhancement and selection module (FESM) is proposed to enhance the single-scale features and select the multi-scale features for reducing the confusing information. During the training, the auxiliary image-level supervision (AIS) is adopted to speed up the convergence and to enhance the feature discrimination of the target defects. The experiment results show that the proposed MSC-DNet reaches the accuracy of 79.4% mAP and 14.1 FPS on NEU-DET dataset, and 71.6% mAP on GC10-DET dataset among all the benchmark methods, which satisfies the quasi-real-time requirement in multiscale defect detection task. • The accurate detection of medium-scale, large-scale defects plays an important role. • A context enhancement module is proposed to extract complementary information. • Image-level supervision is adopted to enhance the feature discrimination. • MSC-DNet achieves 79.4% mAP, 14.1 FPS on NEU-SET set, and 71.6% mAP on GC10-DET set. [ABSTRACT FROM AUTHOR]

This study investigated the interaction among different components of asphalt binders collectively modified with crumb rubber and warm mix additives through thermal gravimetric analysis (TGA) and differential scanning calorimeter (DSC) testing. The modified binders were divided into two parts for analysis purpose, namely the extracted crumb rubber modifier (CRM) by removing all asphalt components and the separated binder system by removing insoluble CRM particles. The main findings of this study include that: (1) the interaction between CRM and base asphalt was a component exchange process, i.e., natural rubber and styrene-butadiene rubber were released from CRM and blended with asphalt molecules; (2) wax-based warm mix additives penetrated into CRM during the mixing process; (3) the incorporation of warm mix additives especially the wax-based additives, promoted the component exchange between rubber and asphalt. [ABSTRACT FROM AUTHOR]

Highlights: [•] We prepared the lightweight aggregate concrete (LWC) containing different mineral admixtures. [•] We studied the carbonization characteristics of LWC. [•] Internal pH of sample with grade II fly ash or steel slag decrease significantly. [•] High volume mineral admixture replacing cement decreases the internal pore solution’s alkalinity of LWC. [•] Effects of pulverized fly ash and granulated blast furnace slag on LWC alkalinity are not obvious. [Copyright &y& Elsevier]

Class imbalance problem commonly exists in multi-label classification (MLC) tasks. It has non-negligible impacts on the classifier performance and has drawn extensive attention in recent years. Borderline oversampling has been widely used in single-label learning as a competitive technique in dealing with class imbalance. Nevertheless, the borderline samples in multi-label data sets (MLDs) have not been studied. Hence, this paper deeply discussed the borderline samples in MLDs and found they have different neighboring relationships with class borders, which makes their roles different in the classifier training. For that, they are divided into two types named the self-borderline samples and the cross-borderline samples. Further, a novel MLDs resampling approach called Multi-Label Borderline Oversampling Technique (MLBOTE) is proposed for multi-label imbalanced learning. MLBOTE identifies three types of seed samples, including interior, self-borderline, and cross-borderline samples, and different oversampling mechanisms are designed for them, respectively. Meanwhile, it regards not only the minority classes but also the classes suffering from one-vs-rest imbalance as those in need of oversampling. Experiments on eight data sets with nine MLC algorithms and three base classifiers are carried out to compare MLBOTE with some state-of-art MLDs resampling techniques. The results show MLBOTE outperforms other methods in various scenarios. • A new borderline oversampling technique for multi-label imbalanced learning. • Defining self-borderline and cross-borderline samples in multi-label data sets. • Handling one-vs-rest imbalance in multi-label imbalanced learning. • Performing competitively in the experiments with C4.5, RBF kernel SVM and linear SVM. [ABSTRACT FROM AUTHOR]

Asphalt rubber (AR) is a sustainable paving material with merits including waste tire consumption, low traffic noise, and enhanced mechanical performance. However, the poor workability and storage stability limited its further application. This study attempted to alleviate these two concerns of AR simultaneously by incorporating heavy bio oil (HBO). To achieve this goal, bio-AR binders with three different mixing sequences were prepared. A series of rheological and chemical tests were conducted. Test results prove that the bio-AR binders exhibited superior rutting and fatigue resistance compared to AR binder. The viscosity values of bio-ARs were closed to AR modified with commercial warm mix additive, which indicates enhancement in workability. Due to the relatively high density of HBO, the density difference between the asphalt liquid phase and crumb rubber in the bio-AR system narrowed, which brought improved storage stability. Among bio-ARs prepared with different mixing sequences, the direct mixing one (ARB) had the most satisfied overall performance. The early incorporation of HBO had limited negative influence on binder performance, but allowed for more energy saving during the bio-AR binder production. Future study will be conducted on performance of bio-AR mixtures and quantitative estimation of its energy saving during the blending and compacting process. [ABSTRACT FROM AUTHOR]

Vertical-axis wind turbines (VAWTs), which are considered an important wind energy conversion device, have recently received renewed attention. Researchers are investigating the pitch angle control method to improve the wind energy utilization efficiency of VAWTs. In this study, ANSYS Fluent is used to investigate the aerodynamic performance of two-dimensional VAWTs. Different pitch angles are investigated using unsteady Reynolds-averaged Navier–Stokes calculations, and turbulence is modeled with the shear stress transport k–ω model. Four quartiles are defined for the azimuthal angle of a rotation period: upwind (45 ° < θ < 135 °), leeward (135 ° < θ < 225 °), downwind (225 ° < θ < 315 °), and windward (315 ° < θ < 45 °). Furthermore, the sliding-mesh technique and a user-defined function are applied to adjust the pitch angle of the wind turbine blades in different azimuthal zones. This study proposes a pitch angle control method that effectively improves the efficiency of VAWTs by comprehensively analyzing the aerodynamic performance of the wind turbine and power coefficient (C p) variation in each zone. The results indicate that an appropriate pitch angle control solution can significantly improve the efficiency of the wind turbine. Compared to a fixed pitch angle of 0°, an approximately 17.0% increase in C p can be achieved and, compared to the optimal fixed pitch angle −1°, an approximately 12.0% increase in C p can be achieved. • Pitch angles of −5° to +4° and the effect on the VAWT power are investigated. • Obtain the optimal pitch angle for each quartile to optimize blade pitch control. • Suitable pitch angle control solution increases the turbine C p by 12.0% compared to ADNI. [ABSTRACT FROM AUTHOR]

Warm mix asphalt (WMA) is an emerging clean production technology that alleviates the energy and environmental concerns of asphalt pavement industry. It is particularly suitable for asphalt rubber (AR) pavement, which provides longer service life and lower road-tyre noise but requires higher production temperature. Evotherm-DAT, a common liquid warm-mix asphalt (WMA) additive, can effectively improve the workability of AR binder, thus reducing the construction temperature of AR pavement. However, the properties of the Evotherm-DAT modified AR binders (Evo-AR) might be affected by the procedure of incorporating the WMA additive into AR, which unfortunately has not been fully studied yet. This study aims to address this issue by characterizing the rheological properties and chemical compositions of the Evo-ARs prepared by the following three procedures: 1) preparing AR first and then blending it with Evotherm-DAT (AR-E), the conventional approach; 2) mixing Evotherm-DAT and rubber first and then incorporating them to asphalt (ER-A); and 3) adding Evotherm-DAT during the mixing process of AR (REA). It was found that AR-E and REA had similar rheological properties. Besides, the workability of ER-A was slightly worse than those of AR-E and REA, but still much superior to that of AR without WMA additive. The anti-rutting performance of ER-A is better than those of AR-E and REA while no obvious difference was observed in the fatigue and low temperature cracking resistance among three types of Evo-ARs. Chemical characterization indicated that the preparation procedures affected the amount of epoxy groups in Evo-ARs, which possibly contributed to the difference in their rheological properties. Overall, it was concluded that incorporating Evotherm-DAT at an early stage has insignificant negative effect on the rheological properties of warm AR binder, but allows for more energy saving and emission reduction during the warm AR binder production process. [ABSTRACT FROM AUTHOR]

This study aims to investigate the rheological and chemical characteristics of asphalt rubber (AR) binders modified with four non-foaming warm mix asphalt (WMA) additives: Evotherm-DAT, Sasobit, conventional paraffin wax, and combined Evotherm-DAT and Sasobit. The main findings of this study include that: (1) all selected WMA additives are effective in enhancing AR’s workability; (2) using combined Evotherm-DAT and Sasobit is not a viable option to reduce the construction temperature of AR mixture; (3) paraffin wax is a potential WMA additive for AR; and (4) WMA additives may affect the dissolution status of crumb rubber in base asphalt. [ABSTRACT FROM AUTHOR]

Modern horizontal-axis wind turbines improve their performance by controlling the blade pitch, which is a well-developed technique in practical use. However, the application of this method in vertical-axis wind turbines (VAWTs) is restricted, due to continuous changes in both the relative wind velocity and angle of attack of the blade. In the study, through an in-depth aerodynamic analysis between the blade pitch and performance of VAWTs, a real-time feedback blade pitch control system is developed, which is based on the real-time flow velocity around the blade. Subsequently, a computational fluid dynamics method with FLUENT is constructed to evaluate the performance of the pitch control system on improving VAWT performance. The calculation method is verified via wind tunnel experimental data. Results indicate that real-time feedback control of the pitch angle increases the average power coefficient in a wide range of tip speed ratios (TSRs). At low TSRs, it significantly improves the starting performance by decreasing the occurrence of vortex shedding and flow separation on the blade surface. At high TSRs, it effectively increases the power coefficient of VAWTs. Additionally, at high TSRs, the power coefficient generally exceeds 0.3 and the maximal power coefficients of the VAWTs corresponds to 0.441. At high TSRs, when compared to the zero pitch, the power coefficient relatively increases by at least 12.7%. • A novel real-time blade pitch control system for VAWTs is developed. • A method evaluates performance of pitch control system in CFD is developed. • This pitch control system effectively improves power coefficient of VAWTs. [ABSTRACT FROM AUTHOR]

Asphalt rubber (AR), which is prepared by blending crumb rubber and bitumen, provides various advantages, including superior rutting resistance, lower road-tire noise and longer service life. However, contractors have expressed concerns regarding its poor storage stability, which in turn limits its wider application. This study aims to address the storage stability concern by incorporating nano-montmorillonite (nanoclay). Three types of nanoclay were dispersed into hot AR binder by high shear blending. The rheological properties of nanoclay-crumb rubber modifier (CRM)-modified bitumen were evaluated through Superpave performance grade (PG) tests and the storage stability was characterized by measuring the difference in softening points or complex moduli at the top and bottom portions of binders after lab-simulated storage. X-ray diffraction (XRD) evaluation was conducted to observe the variation of nanoclay layer gap distance for mechanism investigation. It was found that all selected nanoclays had insignificant effects on workability, rutting and fatigue properties. The layered nanoclay transformed to intercalated or exfoliated structures after interaction with bitumen fractions, providing superior storage stability. Among the three selected nanoclays, pure montmorillonite with Na+ inorganic group, which has an intermediate hydrophilic property and middle layer gap, showed the most obvious effect on enhancing the storage stability of AR. [ABSTRACT FROM AUTHOR]