Li Zha's research while affiliated with Chinese Academy of Sciences and other places

... In this paper, the Aadhaar data analysis is carried out on different distributed computing frameworks mainly MapReduce [1], Hive [3] and Apache Spark [4] on top of Hadoop. ...

... For example, MLPerf [50] is a comprehensive benchmark for measuring ML inference performance across a spectrum of use cases. Architecture-oriented DNN benchmarks [39], [32], [38] target on analyzing the architectural features of DNNs on computing systems of different sizes. MDLBench [39], Embench [32] and AIoTBench [23] are representative benchmarks that characterize the features of different AI models on edge or mobile devices while NNBench-X [57], GNNMark [46] target on acceleration hardware design for different DNNs. ...

... Many large-scale distributed computing organizations that need to store and maintain continuous amounts of data deploy distributed storage systems, such as HDFS [4,5], GFS [2,3] (which were mentioned above), Ceph [12], Microsoft Azure [13,14], Amazon S3 [15], Alluxio [16] etc., which comprise multiple nodes, often organized into groups called racks. Currently, most of these systems write and store large data as blocks of fixed size, which are distributed almost evenly among the system's nodes using random block placement or load balancing policies. ...

... Here, the "Spark+HBase" scheme performs the worst which is mainly because HBase is not a good storage engine for processing complex queries with a large number of data accesses. HBase is mainly designed for point query and some range queries with the help of additional index techniques [17] . Compared with "Spark+Hive", CirroData has a better distributed query execution plan, and the tasks are scheduled in a more load-balanced manner. ...

... For example, when executing an UPDATE statement, MariaDB updates a table's secondary indexes in the same transaction as the table rows [91]. However, in systems that implement asynchronous (lazy) derived state maintenance policies [111,130,135] derived state can become stale with respect to corpus. ...

... [30] models the information propagation as heat diffusion in a high dimensional latent space through which the node's representation is learned. Following this work, [31,32] make modifications to improve the performance. The introduction of the deep learning method makes the use of network topology, temporal order and other features much more convenient, giving rise to a quick shift in designing the model. ...

... Sometimes not all mobile have cameras as well. Some devices can play video and some cannot play video [42]. Quality assurance of mobile apps deals with high-quality apps is produced based on standards and deliver highquality products line for the consumers. ...

... After acquiring data and understanding the data structure, the process which distinguish the standard analytic project and big data project is data split process which applying the concept of map and reduce. In map and reduce concept, the master distributes the idle task to the slave [25]. A large file which is infeasible for standard processing tools is distribute to several files. ...

... Since the stage is set to have research experience of resource discovery, it is imperative for libraries to provide information with tailor-made expansive representation of library collections and relevant content. Finding a way to combine the best of the just-in-case and Library automation just-in-time technologies using the pre-harvested, pre-massaged and pre-indexed approach of search engines to meet changing user expectations is the current challenge for librarians and their vendors (Walker, 2009). According to a report by the Research Information Network (2008) on building research tools from the physical artefacts and resources of museums, "technological developments offer opportunities for cross searching, for making records findable by Google and other search engines, for linking to associated documentation, and for integrating museum catalogues with other resources, such as library catalogues. ...

... The research of Grid-Cloud based IT platform mainly includes three sub-fields, system architecture, software mechanism and resource allocation. A great deal of existing research concerning software mechanism that will be necessary to bring Grid and Cloud to fruition is underway9101112131415161718. Thereby, we are able to abstract away implementation details and focus on the scalable service system architecture and the effective Grid-Cloud resource allocation. ...