两种面向推荐系统的数据压缩方法

今日头条的服务器每天都会产生规模庞大的训练数据,为方便进行训练,这些数据都具有特定的格式和分布特征。使用不同类型的通用压缩算法(字典类型及非字典类型)进行测试,发现单独任何一种算法都无法在满足业务需求(速率需求和CPU占比等)的同时获得较为可观的压缩比。针对今日头条的训练数据,提出了分段聚类压缩和Hash recoding压缩两种策略。实验结果表明,分段聚类压缩方式在更好地保证了压缩率的同时提高了压缩速率的目的;Hash recoding压缩方式则更好地达到了以少量压缩速率的损失换取更可观的压缩率的目的。分段聚类方式搭配Gzip压缩算法的组合能使压缩速率提高300%以上;Hash recoding方式匹配Snappy压缩算法能使压缩率缩小50%以上。根据实际需求,不论选择哪种策略,对于降低今日头条的运营成本,提升业务处理的效率,提供更好的用户体验,都有一定意义。     

An efficient and noise resistive selective image encryption scheme for gray images based on chaotic maps and DNA complementary rules

A novel image encryption algorithm in streaming mode is proposed which exhaustively employs an entire set of DNA complementary rules alongwith one dimensional chaotic maps. The proposed algorithm is highly efficient due to encrypting the subset of digital image which contains 92.125 % of information. DNA addition operation is carried out on this MSB part. The core idea of the proposed scheme is to scramble the whole image by means of piecewise linear chaotic map (PWLCM) followed by decomposition of image into most significant bits (MSB) and least significant bits (LSB). The logistic sequence is XORed with the decoded MSB and LSB parts separately and finally these two parts are combined to get the ciphered image. The parameters for PWLCM, logistic map and selection of different DNA rules for encoding and decoding of both parts of an image are derived from 128-bit MD5 hash of the plain image. Simulated experimental results in terms of quantitative and qualitative ways prove the encryption quality. Efficiency and robustness against different noises make the proposed cipher a good candidate for real time applications.     

基于CCD技术的外观纸病检测系统

设计了CCD在线外观纸病检测系统,介绍了该系统的硬件组成情况和系统的工作原理、技术指标和软件设计,重点阐述了外观纸病检测系统中CCD技术的优势。     

基于北斗的高速摄像机时间同步器设计

为了满足民用飞机在飞行试验中高速摄像机对时间同步的需求,基于国产北斗卫星导航系统设计了具有精确输出IRIG-B(AC)码的时间同步器,用于机载高速摄像机的精确时间同步。以FPGA为核心,使用直接数字频率合成技术实现了IRIG-B(AC)码的产生。实验结果表明,该同步器时间同步精度满足要求,可以实现对高速摄像机进行时间同步的功能。     

Evolutionary variational inequality with long-term memory and applications to economic networks

The aim of this paper is to present the relation between an evolutionary variational inequality with long-term memory and Lagrange multipliers. More precisely, we study the oligopolistic market equilibrium problem in which the profit function depends also on previous events of the market by means of a long-term memory which takes into account the previous states of the equilibrium. Moreover, thanks to the variational formulation, we are able to show existence and regularity results for equilibrium solutions. Then, we apply the infinite dimensional duality theory through which we obtain the existence of Lagrange multipliers which are great utility in order to understand the behaviour of the market. Finally, an example is provided, which allows to analyse the influence of the long-term memory on the equilibrium solution.     

The Rise of Twitter in the Political Campaign: Searching for Intermedia Agenda‐Setting Effects in the Presidential Primary

Questions exist over the extent to which social media content may bypass, follow, or attract the attention of traditional media. This study sheds light on such dynamics by examining intermedia agenda‐setting effects among the Twitter feeds of the 2012 presidential primary candidates, Twitter feeds of the Republican and Democratic parties, and articles published in the nation's top newspapers. Daily issue frequencies within media were analyzed using time series analysis. A symbiotic relationship was found between agendas in Twitter posts and traditional news, with varying levels of intensity and differential time lags by issue. While traditional media follow candidates on certain topics, on others they are able to predict the political agenda on Twitter.