Search results

Recently, with the rapid development of network, communications and computer technology, privacy preserving data mining (PPDM) has become an increasingly important research in the field of data mining. In distributed environment, how to protect data privacy while doing data mining jobs from a large number of distributed data is more far-researching. This paper describes current research of PPDM at home and abroad. Then it puts emphasis on classifying the typical uses and algorithms of PPDM in distributed environment, and summarizing their advantages and disadvantages. Furthermore, it points out the future research directions in the field.

Online portfolio selection sequentially allocates wealth among a set of assets and aims to maximize the investor’s cumulative return in the long run. Various existing algorithms in the finance and accounting area adopt an indirect approach to exploit one asset characteristic through the channel of assets’ expected return and thus cannot fully leverage the power of various asset characteristics found in the literature. This study aims to propose new algorithms to overcome this issue to enhance investment performance.

We propose a parameterized portfolio selection (PPS) framework, which directly incorporates multiple asset characteristics into portfolio weights. This framework can update parameters timely based on final performance without intermediate steps and produce efficient portfolios. We further append L1 regularization to constrain the number of active asset characteristics. Solving the PPS formulation numerically, we design two online portfolio selection (OLPS) algorithms via gradient descent and alternating direction method of multipliers.

Empirical results on five real market datasets show that the proposed algorithms outperform the state of the arts in cumulative returns, Sharpe ratios, winning ratios, etc. Besides, short-term characteristics are more important than long-term characteristics, and the highest return category is the most important characteristic to improve portfolio performance.

The proposed PPS algorithms are new end-to-end online learning approaches, which directly optimize portfolios by asset characteristics. Such approaches thus differ from existing studies, which first predict returns and then optimize portfolios. This paper provides a new algorithmic framework for investors’ OLPS.

Tingting Qi's chapter titled, “Moving toward Decentralization? Changing Education Governance in China After 1985,” provides the historical and policy context for the volume. This chapter integrates the post-1978 Chinese educational reforms into the socioeconomic context of China. The special contribution of this chapter is that it explores the complexity of educational decentralization in China through an in-depth analysis of changes in education finance, administration, and curriculum. Qi reviews prior studies, government documents, laws, and regulations related to Chinese education reform since 1978 within the context of education decentralization in China. Qi also demonstrates that China's educational policy reforms are moving China toward “centralized decentralization” because decentralization is driven by a common, centralized national goal of economic modernization. The chapter presents evidence that “centralized decentralization” is a strategic maneuver that maintains centralized control while providing the reform legitimacy of decentralization. By focusing on decentralization as the core of Chinese educational policy reforms, this chapter situates the following chapters within the social, cultural, and political context of post-1978 China.

This study aims to present a more accurate lifetime prediction model considering solder chemical composition.

Thermal cycling and standard creep tests as well as finite element simulation were used.

The study found lower error in the solder joint lifetime evaluation. The higher the Ag content is, the higher the lifetime is achieved.

It is confirmed.