Papers by Keyword: Maximum Entropy

Abstract: Crop disease leaf image segmentation is a key step in crop disease recognition. In the paper, a segmentation method of crop disease leaf image is proposed to segment leaf image with non-uniform illumination based on maximum entropy and genetic algorithm (GA). The information entropy is regarded as the fitness function of GA, the maximum entropy as convergence criterion of GA. After genetic operation, the optimal threshold is obtained to segment the image of disease leaf. The experimental results of the maize disease leaf image show that the proposed method can select the threshold automatically and efficiently, and has an advantage over the other three algorithms, and also can reserve the main spot features of the original disease leaf image.

Abstract: This paper proposes a novel multi-view learning framework which leverages the information contained in pseudo-labeled images to improve the prediction performance of image classification using multiple views of an image. In the training process, labeled images are first adopted to train view-specific classifiers independently using uncorrelated and sufficient views, and each view-specific classifier is then iteratively re-trained using initial labeled samples and additional pseudo-labeled samples based on a measure of confidence. In the classification process, the maximum entropy principle is utilized to assign appropriate category label to each unlabeled image using optimally trained view-specific classifiers. Experimental results on a general-purpose image database demonstrate the effectiveness and efficiency of the proposed multi-view semi-supervised scheme.

Abstract: Process quality evaluation plays a significant role in ensuring product quality and improving customer satisfaction. Currently, the research on process quality evaluation lacks consideration of interactive coupling between product processing parameters and equipment reliability. With using the analysis of physics of failure, the relation model between the products quality and the processing equipment reliability was established. Besides, based on Maximum Entropy Principle (MEP), an evaluation method of process quality was provided. Finally, we verified the feasibility of the method through the combination in an application case.

Abstract: With the international economic and trade exchanges, biological invasion has become an important factor in the threat of biological diversity and ecological environment. In order to prevent and control exotic species invasion, using of GIS-based geospatial technology to the potential distribution of invasive species risk assessment can provide the basis for ecological management of invasive alien species better, thus reducing the economic losses due to invasive alien species caused to our country. To this point, the study had done a secondary development based on ArcGIS Engine component. It used niche model to analysis and study potential suitable distribution of alien pests in China, conduct a risk assessment to them. The receiver operating characteristic curve was applied to access the prediction accuracy Analyzed the impact of the distribution of invasive alien species` main environmental variables. It provided a theoretical basis for ecological management of invasive alien species.

Abstract: In order to cope with the increasing size of the training corpus and adapt to the requirements of incremental learning, this paper introduces a feature selection algorithm of maximum entropy model into the research of Chinese word segmentation technology, designs and implements a Chinese word segmentation system based on incremental learning. The experimental results show that the system gradually improves the segmentation accuracy in the incremental learning process which without wasting time to restudy.

Abstract: To solve the difficulty of equipment testability verification experiment sample selection, a sample selection method of testability verification experiment was presented. The failure-test dependency matrix and its extended matrix, failure modes function equivalence set and failure modes test equivalence set were analyzed, on the basis, failure modes sample equivalence set was proposed. Meanwhile, importance characteristic was analyzed and maximum entropy was solved, and the sample selection method was presented, its process was obtained. Moreover, the method was comparatively analyzed with actual methods in the experiment. The results show that application the method in paper, can reduce sample number, save test expenses, and meet the test requirement of adequacy, so the method in paper is proved effective and feasible.

Abstract: According to the limitations of the shooting environment, captured image exist the phenomenon of image blurring and noise. This paper proposes that the improved maximum entropy method recovery blurred image which acquire in aerial. Finally, according to the first order Markoff theory to evaluate the quality of the processed image, the results show that maximum entropy image restoration method compared to the conventional approach increase image clarity and details more better.

Abstract: It is difficult to assess the impact of voltage sags on sensitive equipment because of the complex uncertainties involved in voltage tolerance capability. Determining the distribution regularity of voltage tolerance capability is an important step in voltage sag sensitivity assessment. Computers’ voltage tolerance capability are obtained through laboratory tests. The point-value distribution function of voltage tolerance capability is calculated by Maximum Entropy Principle and then the interval probability distribution function is derived based on confidence interval, so as to derive an interval probability of equipment failure probability. The proposed model is finally verified in the IEEE-30 bus system. The results are compared with that of stochastic estimation method, showing that the proposed method can reflect actual conditions better and obtain accurate result.

Abstract: Recently, sentiment analysis of text is becoming a hotspot in the study of natural language processing, which has drawn interesting attention due to its research value and extensive applications. This paper introduces a smart sentiment analysis system, which is to satisfy three aspects of sentiment analysis requirement. These are Chinese sentiment word recognition and analysis, sentiment related element extraction and text orientation analysis. Promising results and analysis are presented at the end of this paper.

Abstract: This paper is mainly about the experiment research of oil monitoring for a kind of special heavy vehicle’s transmission system. The wear elements in lubrication oil of a heavy vehicle’s transmission system were analyzed by the spectroscopic analysis method, and gear’s wear status was studied. Probability density function and probability distribution function of wear information about the elements in lubricating oil were calculated by the maximum entropy theory. The running cut-off point is obtained for diagnostic criteria of the vehicle’s failure, which provided evidence for monitoring and judging the reliability of the vehicle’s wear.