Minimalization (also called miniaturization) refers to the removal of all non functional and necessary source code characters (such as white space, line feed, annotation, and some block recognizers) without affecting the function of the programming language, because although they help to improve the readability of the source code, they are not necessary in actual operation. In the category of programming language, it refers to the removal of all non functional and necessary source code characters without affecting the function.
Minimalization (also called miniaturization) refers to the removal of all non functional and necessary source code characters (such as white space, line feed, annotation, and some block recognizers) without affecting the function of the programming language, because although they help to improve the readability of the source code, they are not necessary in actual operation.
基于 Hadoop 的个性化推荐系统在电子商务的应用
Application of personalized recommendation system based on Hadoop in E-commerce
随着电子商务推荐系统的规模越来越大，运算量也随之增加，运算所需要的资源和时间耗费明显增大。Hadoop 平台给这个瓶颈带来的曙光，Hadoop 所具有的并行处理、容错处理、本地化计算、负载均衡等优点，极大的简化了并行程度设计的难度，利用该平台处理海量的电子商务数据，通过 MapReduce 优化程度处理流程，验证算法在 Hadoop 集群的加速比和扩展性等方面取得了较好的效果。
With the increasing scale of e-commerce recommendation system, the amount of computation increases, and the resource and time consumption of computation increases significantly. Hadoop platform brings the dawn to this bottleneck. The advantages of Hadoop, such as parallel processing, fault-tolerant processing, localized computing, load balancing, greatly simplify the difficulty of parallel degree design. The platform is used to process massive e-commerce data, and MapReduce is used to optimize the degree processing process to verify the application of the algorithm in Hadoop The cluster has achieved good results in speedup and scalability.
Research on target recognition method of rehabilitation robot based on image local features
提出了基于快速 SIFT 算法的目标识别方法。SIFT 算法存在的主要不足是高维数的 SIFT 特征描述符计算复杂，造成算法实时性较差。为简化算法计算复杂程度，同时保证不损失正确匹配特征，首先构建目标图像尺度空间，提取 SIFT 特征点时将其按大小分类，然后扩展 SIFT 角度属性，由 SIFT 特征点子区域方向直方图计算得到 4 个新角度，代表特征点方向信息，最后在特征匹配时，根据 SIFT 特征点角度信息以及大小来限制特征点匹配范围，简化算法复杂程度，得到快速 SIFT 算法。实验结果表明，应用快速 SIFT 算法有效提高了特征匹配效率。
A target recognition method based on fast SIFT algorithm is proposed. The main disadvantage of SIFT algorithm is that the computation of high-dimensional SIFT feature descriptor is complex, resulting in poor real-time performance of the algorithm. In order to simplify the computational complexity of the algorithm and ensure that the correct matching features are not lost, the scale space of the target image is constructed firstly, and the SIFT feature points are classified according to their size when they are extracted. Then, the SIFT angle attribute is extended, and four new angles are calculated from the direction histogram of the SIFT feature sub region, representing the direction information of the feature points. Finally, in feature matching, the SIFT angle attribute is used to calculate the direction information of the feature points The angle information and size of feature points are used to limit the matching range of feature points, simplify the complexity of the algorithm, and get a fast SIFT algorithm. The experimental results show that the fast SIFT algorithm can effectively improve the efficiency of feature matching.