Search

[Purpose/Significance] At present, most of the training models used in the research of news classification are non-active learning. There are common problems about these models, including data cannot be labeled immediately and the labeling cost is too high, which also hinders the analysis of agricultural news. Especially because of the explosive growth of news data in the network era, it is more difficult to label data, train supervised text classification models, and screen relevant news in the field of agriculture from diversified online news sources. In order to solve this problem, the most commonly used pool based active learning or deep active learning technique is used to select more valuable and representative data from unlabeled data for manual labeling, and construct labeled data sets to improve the efficiency and effect of news classification and agricultural news mining. [Method/Process] The commonly used machine learning models for text classification, such as random forest classifier, polynomial naive Bayes classifier and logistic regression classifier, were combined with the active learning method with the lowest confidence to analyze the effect, and the BERT model was combined with the three sampling strategies of discriminative active learning, deep Bayes active learning and lowest confidence for deep active learning training. On the news corpus of 19 847 samples crawled and cleaned by crawler technology from Sina and other news websites, aiming at screening agricultural related news from diversified news samples of various topics, the iterative experiment of adding 30 samples per round was tested to check the improvement effect of F1 score under various method combinations with the increase of the number of annotation. In addition, the representativeness and diversity of the samples selected by the sampling function of each method in the deep active learning method of the BERT model were compared, so as to understand the characteristics of each strategy and provide inspiration for the selection and improvement of Al strategy in the future. In addition, this paper also analyzed how much labeling cost can be saved by using the proposed method. [Results/Conclusions] When comparing a variety of machine learning models, it is found that although the gradient boosting tree and support vector machine classifier have high accuracy, they are not suitable for active learning because of their low efficiency in text data processing of large-scale high-dimensional data. After combining other machine learning models and the BERT model and training text models with the corresponding active learning or deep active learning methods, it is found that the application of active learning method can significantly improve the training process of each model. Among them, the BERT model, combined with discriminative active learning sampling function, has the best news text classification effect and the lowest annotation data requirements. The representativeness and diversity of the samples selected by discriminative active learning sampling function are also the highest, which explains the source of the advantages of this method. It can also be found that for the same task model, the higher the accuracy of classification is required, and the active learning method can save more annotation cost than non-active learning.

This paper presents a hybrid linear actuator using screw clamp operation principle. The actuator mainly consists of a hollow electromagnetic torque motor located between two clamping nuts, two hollow cylindrical shaped piezoelectric stacks symmetrically configured at two ends of the actuator and a feed-screw (also considered as the mover of the actuator) assembled throughout all the parts. The torque motor is symmetrically connected to two clamping nuts via two torsion coupling springs located at either end of the motor spindle. Two piezoelectric stacks can work independently to propel the opposing loads, which effectively take advantage of the anti-compression and non-tensile characteristics of piezoelectric element. The special feature of the actuator is the screw clamp mechanism, the operation of which involves intermittent rotation of two nuts (driven by the torque motor) on a feed-screw to achieve the bi-direction piezoelectric motion accumulation. Furthermore, the application of feed-screw could decrease the actuator's sensitivity to wear, in order to realize a rigid self-locking and thus ensure the actuator's holding capacity. A prototype was fabricated and the experimental results show that the no-load speed, maximum thrust, and peak power of the actuator were 20 mm/s, 280 N, and 1.54 W, respectively. [ABSTRACT FROM AUTHOR]

Purpose This study systematically investigated the effect of the binary rare earth oxide of La2O3 and Sm2O3 on the properties of the Al2O3/TiO2 (AT) coating, including phase transform, wear behavior, etc.Design/methodology/approach AT coatings mixed with different components of binary rare earth oxides of La2O3 and Sm2O3 are prepared by atmospheric plasma spraying. The adhesion strength, micro-hardness, phase transition and tribological behavior of coatings are systematically investigated.Findings The X-ray diffraction (XRD) analysis shows that phase transformation is obvious after spraying, and a-Al2O3 is almost translated into γ-Al2O3 when La2O3 and Sm2O3 are doped together. Meanwhile, solid solution generated between rare earth oxide and Al2O3/TiO2 coatings results in disappearance of TiO2 and rare earth oxide phase. The photos under the scanning electron microscope (SEM) indicate that binary rare earth oxide could increase the melting degree of powder and decrease porosity of coatings.The increasing of Sm2O3 rarely affect micro-hardness and adhesion strength, and the coating with 4 per cent Sm2O3 and 1 per cent La2O3 exhibits the best wear resistance and lowest friction coefficient among all the samples.Originality/value AT coatings mixed with different components of binary rare earth oxide of La2O3 and Sm2O3 are prepared by atmospheric plasma spraying. Binary rare earth oxide could increase the melting degree of powder and decrease porosity of AT coatings. [ABSTRACT FROM AUTHOR]

To solve the problems that double-mode-drive-type moving ultrasonic motors (USMs) need to equalize two natural frequencies for driving and that single-mode-drive-type USMs either only move uni-directionally or have many difficulties in realizing bidirectional movement, a single-mode-drive-type bidirectional linearly moving USM composed of an inclined slider and a V-shaped stator is presented. The slider is obliquely installed with the stator which works under the in-plane symmetric vibration mode or the in-plane unsymmetrical vibration mode. When the working mode of stator is changed, the inclined direction of the trajectory of driving-foot is changed, so the stator can move the slider bidirectionally. On the basis of analyzing the working theory and the design principles of the motor, the obliquity range of the inclined slider which can make the motor running normally is derived. Finally, a prototype is fabricated and its vibration mode and mechanical characteristics are tested. The experimental results show that under the conditions that the obliquity angle of slider is 35°, the peak-to-peak voltage is 600 V and the pre-pressure is 65 N, when the motor is working under the in-plane symmetric vibration mode, the motor offers a maximal driving force of 14 N, and the maximal velocity with no load is 180 mm/s; when the motor is working under the in-plane unsymmetrical vibration mode, the motor offers a maximal driving force of 24 N, and the maximal velocity with no load is 125 mm/s. [ABSTRACT FROM AUTHOR]