Learning from imbalanced data is a challenging task in the machine learning field, as with this type of data, many traditional supervised learning algorithms tend to focus more on the majority class ...while damaging the interests of the minority class. Stacking ensemble, which formulates an ensemble by using a meta-learner to combine the predictions of multiple base classifiers, has been used for solving class imbalance learning issues. Specifically, in the context of class imbalance learning, a stacking ensemble learning algorithm is generally considered to combine with a specific sampling algorithm. Such an operation, however, might suffer from suboptimization problems as only using a sampling strategy may make it difficult to acquire diverse enough features. In addition, we also note that using all of these features may damage the meta-learner as there may exist noisy and redundant features. To address these problems, we have proposed a novel stacking ensemble learning algorithm named MSFSS, which divides the learning procedure into two phases. The first stage combined multiple sampling algorithms and multiple supervised learning approaches to construct meta feature space by means of cross combination. The adoption of this strategy satisfied the diversity of the stacking ensemble. The second phase adopted the whale optimization algorithm (WOA) to select the optimal sub-feature combination from the meta feature space, which further improved the quality of the features. Finally, a linear regression classifier was trained as the meta learner to conduct the final prediction. Experimental results on 40 benchmarked imbalanced datasets showed that the proposed MSFSS algorithm significantly outperformed several popular and state-of-the-art class imbalance ensemble learning algorithms. Specifically, the MSFSS acquired the best results in terms of the F-measure metric on 27 datasets and the best results in terms of the G-mean metric on 26 datasets, out of 40 datasets. Although it required consuming more time than several other competitors, the increment of the running time was acceptable. The experimental results indicated the effectiveness and superiority of the proposed MSFSS algorithm.
Heterogeneous cross-project defect prediction (HCPDP) aims to learn a prediction model from a heterogeneous source project and then apply the model to a target project. Existing HCPDP works mapped ...the data of the source and target projects in a common space. However, the pre-defined forms of mapping methods often limit prediction performance and it is difficult to measure the distance between two data instances from different feature spaces. This paper introduced optimal transport (OT) theory for the first time to build the relationship between source and target data distributions, and two prediction algorithms were proposed based on OT theory. In particular, an algorithm based on the Gromov-Wasserstein (EGW) entropic discrepancy was developed to perform the HCPDP model. The proposed EGW model measures the distance between two metric spaces by learning an optimal transfer matrix with the minimum data transfer cost and avoids measuring the distance of two instances of different feature spaces. Then, to improve EGW performance, an EGW+ transport algorithm based on EGW was developed by integrating target labels. Experimental results showed the effectiveness of EGW and EGW+ methods, and proved that our methods can support developers to find the defects in the early phase of software development.
With the rapid development of Internet information technology, the advantages of social media in terms of speed, content, form, and effect of communication are becoming increasingly significant. In ...recent years, more and more researchers have paid attention to the special value and role of social media tools in disaster information emergency management. Weibo is the most widely used Chinese social media tool. To effectively mine and apply the emergency function of disaster situation microblogs, a disaster situation information discovery and collection system capable of online incremental identification and collection are constructed for massive and disordered disaster microblog text streams. First, based on the deep learning- trained word vector model and a large-scale corpus, an unsupervised short-text feature representation method of disaster situation Weibo information is developed. According to the experimental results of the feature combination test and the training set scale test, the SVM algorithm was selected for disaster microblog information classification, which realized effective identification of disaster situation micro-bloggings. Then, the temporal information similarity and geographic information similarity are used to improve the single text similarity algorithm, and a Chinese disaster event online real-time detection model is constructed. Furthermore, the disaster-affected areas can be achieved in real-time based on the detection results. By crawling and classifying the micro-bloggings from the disaster-affected areas, it is possible to realize the incremental identification and collection of online disaster situation Weibo information. Finally, the empirical analysis of disaster events such as the "Leshan Earthquake" shows that the real- time intelligent identification and collection system for disaster situation Weibo micro-bloggings developed in this paper can obtain large-scale and useful data for disaster emergency management, which proving that this system is effective and efficient.
This study proposes a novel incremental learning algorithm called distribution matching ensemble (DME) in context of adaptive weighted ensemble learning. In particular, DME estimates the distribution ...of each received data block by Gaussian mixture model (GMM) and reserves the corresponding distribution information, as well it maintains a group of classifiers in a buffer. When we receive a new data block which is required to be predicted, the similarity between its distribution and each reserved distribution will be calculated by Kullback-Leibler (KL) divergence, and then the similarities can be used to guide the weight assignment of each corresponding classifier to further make adaptive ensemble decision. DME gets rid of the underlying hypothesis that the most recent labeled data block always has the most similar distribution with the current unlabeled data block. In addition, to avoid infinite extension of ensemble buffer during incremental learning, we also develop two dynamic classifier update rules. Experiments results on some synthetic and real-world streaming datasets show that the proposed DME algorithm is able to track and adapt to various types of concept drift just in time. Especially, on data stream with frequent reoccurring drifts, the DME significantly outperforms to several state-of-the-art algorithms, indicating its superiority.
The pressure–viscosity coefficient (the
α
value), which represents the variation of viscosity as a function of pressure, is an important parameter for elastohydrodynamic lubrication analyses. The ...properties of hydrocarbons in the C
20
–C
40
mass range are of fundamental importance as they are basic constituents of synthetic- and mineral-based lubricant stocks. The conventional acquisition of the
α
value requires preparation of lubricant samples and experimental testing by means of a high-pressure viscometer. In this paper, we present a method to obtain the
α
value of a typical base oil (1-Decene trimer) based solely on the molecular dynamics simulations. Non-equilibrium molecular dynamics (NEMD) simulations were performed to calculate the shear viscosity of the lubricant at various temperatures and pressures up to 1 GPa. Elevated temperatures and time–temperature superposition (TTS)-based extrapolations were applied to further extend the ability of the NEMD simulations, and the rotational relaxation time was calculated and used to determine the validity of the NEMD calculations. The
α
value at 100 °C was calculated and compared with experimental results. Effectiveness of the extrapolation was evaluated with a 95 % confidence interval.
China, where coal is the dominant energy resource, accounted for 50.5% of the world's coal consumption-or 1906.7 million tons of oil equivalent-during 2018. As a major source of volatile organic ...compounds (VOCs), coal utilization also led to high national emissions of these pollutants. In this study, we investigated the profiles; benzene, toluene, ethylbenzene, and xylene (BTEX) ratios; ozone formation potential (OFP); and secondary organic aerosol (SOA) formation potential of VOCs generated by coal-utilizing steel plants, power plants, coking plants, and residential stoves in northern China. Among the detected VOCs, the results identified 1-butene as the most abundant species for both the power plants (36.7%) and the residential stoves (41.7%) as well as a significant contributor at the steel plants (7.3%), and alkenes, alkanes, and aromatics composed the largest groups for the power plants (42.0%) and residential stoves (60.2%); steel plants (59.2%); and coking plants (66.1%), respectively. Additionally, the VOC profiles for power plants employing the same coal source or combustion technology exhibited strong similarities, although the BTEX ratios varied more between plants using different coal sources than those using different combustion technologies. Finally, alkenes were primarily responsible for the ozone formation (73.1%, 59.0%, and 87.6% for the power plants, steel plants, and residential stoves, respectively), whereas aromatics were primarily responsible for the SOA formation (more than 94.0% for all four coal-utilizing sources).
Obvious differences are observed between simulation and experimental results for series-connected insulated-gate bipolar transistors (IGBTs) using current IGBT models. Here, the cause of these errors ...is analyzed in detail. A physical model based on more effective assumptions for a 2-D structure is proposed in this paper. The relationship between carrier concentration and lifetime is considered in the model in order to achieve an accurate description for excess carrier distribution. Testing was performed in a buck converter using series-connected non-punch-through (NPT) planar-gate 6500 V/600 A high-voltage IGBTs (HVIGBTs) at various bus voltages using an asynchronous control signal. The accuracy of HVIGBT transient model is verified by comparing experimental and simulation results in buck converters using two and three series-connected IGBTs. The function of the RC snubber circuit is also evaluated using the proposed model.
During the operation of boom sprayers in the field, it is crucial to ensure that the entire boom is maintained at an optimal height relative to the ground or crop canopy. Active suspension is usually ...used to adjust the height. A control system for double-pendulum active suspension was developed in this paper. The control system consisted of a main control node, two distance measurement nodes, a vehicle inclination detection node, and an execution node. Communication between nodes was carried out using a CAN bus. The hardware was selected, and the interface circuits of the sensors and the actuator were designed. The transfer functions of the active suspension and electric linear actuator were established. In order to enhance the efficiency of the control system, the particle swarm optimization (PSO) algorithm was employed to optimize the initial parameters of the fuzzy PID controller. The simulation results demonstrated that the PSO-based fuzzy PID controller exhibited improvements in terms of reduced overshoot and decreased settling time when compared to conventional PID and fuzzy PID controllers. The experimental results showed that the active suspension system equipped with the control system could effectively isolate high-frequency disturbances and follow low-frequency ground undulations, meeting the operational requirements.
In video coding, temporal correlations between pictures are utilized by short-term and long-term reference, which are limited inside a random access segment and cannot cross random access points ...(RAPs). To improve coding efficiency, correlations across RAPs are exploited by some novel video coding schemes. The cross-random-access prediction structure results in alterable dependency between pictures, which is affected by random access and different from the fixed dependency in conventional video coding standard. However, current media streaming scheme, such as Dynamic Adaptive Streaming over HTTP (DASH), cannot describe dependency across RAPs. This paper introduces the cross-random-access dependency description in DASH syntax. With dependency information, client can download segments in proper temporal order but may re-download segments. This paper further improves downloading management in client. Experiments show that improved DASH system transmitting stream with cross-random-access prediction structure can save 22.1% on maximum and 14.3% on average transmission bits in contrast to conventional DASH system.
The insulated gate bipolar transistor (IGBT) physical models are studied in details. The difference between low-voltage IGBT (LVIGBT) and high-voltage IGBT (HVIGBT) is analyzed and the shortage of ...the LVIGBT model used for HVIGBT is discussed. The physical model considering the effect of carrier concentrate on excess carrier lifetime is established for HVIGBT. The HVIGBT transient model is presented with different excess carrier lifetime in the base. The new description of steady-state U-I characteristics is also obtained with numerical method. The test experiment was performed in a Buck converter using 6500 V HVIGBT with different bus voltage and load current. The accuracy of the transient model of HVIGBT is verified by experiment and simulation results. The verification of some key parameters to describe the external characteristics is also given in the paper.
PDF
