The polarity shift problem is amajor factor that affects classification performance ofmachinelearning-based sentiment analysis systems. In this paper, we propose a three-stage cascade model to address the polarity shift problem in the context of document-level sentiment classification. We first split each document into a set of subsentences and build a hybrid model that employs rules and statis...

This paper describes a new approach for precise image shift detection between two images. The approach is based on an analysis of the phase difference spectrum between two images. Image shifts are directly evaluated from the phase difference spectrum without Fourier inversion. The cross s~ectrum is utilized as the weight parameter distribution. Synthetic and real images of typical image pattern...

The impact-based grain flow sensor is widely used in the combine harvester yield monitor systems to measure the crop yield .but its accuracy wrecked by false signal of excessive natural vibration. A step response experiment was carried to identify the math function of the sensor. Experiment found that the math function of the sensor was a second-order vibration system, its Damping Ratio was 0.0...