[1] This paper, by using the data of Cluster, TC-1, GOES, and eight ground stations on 22 October 2004, studied the characteristics of low-latitude Pi2s generated by an earthward bursty bulk flow (BBF) in the near-Earth tail plasma sheet. The BBF excited simultaneously two distinct classes of Pi2s: one is long-period Pi2 (90-130 s) and the other is short-period Pi2 ($50 s). The long-period Pi2 is transient response type Pi2 associated with field-aligned current produced by the braking of BBFs. The spectrum analysis show that the amplitude spectrum peak of long-period Pi2 increases with increasing latitude, indicating that the source is at higher latitudes. The time delay for the propagation of Alfven waves from Cluster to the Earth is very close to the time difference between the onset time of the BBFs at Cluster and the starting time of the long-period Pi2 on the ground. The short-period Pi2 is a global cavity mode since the Pi2s in H components at eight stations have almost the same starting time, same oscillation period, and same waveform, which are all typical characteristics of cavity mode. The amplitude spectrum peak of short-period Pi2 at NCK (N42.7) is larger than those at higher-station UPS (N56.5) and lower-station CST (N40.8). The polarization analysis at three lower-latitude stations shows that the polarization underwent two reversals. The major axis of the polarization ellipse points to approximately the north, indicating that the short-period Pi2s are not excited by nightside current system. TC-1 observed transverse mode Pi2s. Its period is almost identical with the periods of Pi2 on the ground, indicating they belong to the same wave.

Naive Bayes estimator is widely used in text classification problems. However, it doesn't perform well with smallsize training dataset. We propose a new method based on Naive Bayes estimator to solve this problem. A correlation factor is introduced to incorporate the correlation among different classes. Experimental results show that our estimator achieves a better accuracy compared with traditional Naive Bayes in real world data.

Credit is an important means to promote economic development, while green credit is conducive to the sustainable development of industry. This paper aims to build a multiple linear regression model and a dynamic panel data GMM estimation model to analyze the important factors that affect the optimization of the industrial structure. We then use an analytic hierarchy process to explore the relationship between green credit and industrial optimization. We compare this with the optimization rate of the industrial structure according to real data, and then obtain the effectiveness of the hierarchical analysis of the three major industries in the eastern, central and western regions. Finally, neural networks are used to forecast the total amount and distribution of green credit in 2021. The final results show that there are regional and industrial differences in the influence of green credit on industrial structure optimization, and in the process of using green credit to promote the optimization and upgrading of industrial structure.

Minimum-variance portfolio optimizations rely on accurate covariance estimator to obtain optimal portfolios. However, it usually suffers from large error from sample covariance matrix when the sample size n is not significantly larger than the number of assets p. We analyze the random matrix aspects of portfolio optimization and identify the order of errors in sample optimal portfolio weight and show portfolio risk are underestimated when using samples. We also provide LoCoV (low dimension covariance voting) algorithm to reduce error inherited from random samples. From various experiments, LoCoV is shown to outperform the classical method by a large margin.

We propose a new approach to address the text classification problems when learning with partial labels is beneficial. Instead of offering each training sample a set of candidate labels, we assign negativeoriented labels to the ambiguous training examples if they are unlikely fall into certain classes. We construct our new maximum likelihood estimators with self-correction property, and prove that under some conditions, our estimators converge faster. Also we discuss the advantages of applying one of our estimator to a fully supervised learning problem. The proposed method has potential applicability in many areas, such as crowd-sourcing, natural language processing and medical image analysis.