A Novel Strategy of Non-uniform Sampling Using in Parameter Estimation Based on Fractional Fourier Transform

Tian, Xinluan; Bao, Dan; Feng, Pingyun

doi:10.1109/auteee48671.2019.9033230

Cited by 1 publication

(1 citation statement)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Within this example, there are many important pieces. First, the initial number of records is put in the 'reservoir,' the size of the planned random sample being the same (13). The remainder of the data set is sequentially scanned and the random index from 0 to number of records is generated for each read value.…”

Section: Uniform Random Samplingmentioning

confidence: 99%

Faster as Well as Early Measurements from Big Data Predictive Analytics Model

et al. 2022

View full text Add to dashboard Cite

Data processing in large scale system and analyses is the key problems of today's distributed systems in real-time and close to real time. These systems should be able, while meeting these constraints, to process large-scale data inputs, to expand to terabytes or larger. Volume of processing the data over the systems of large scale becomes too large. These systems use parallelization of the data and similar techniques to increase performance. The standards are increasing, though: data analyses are expected almost in real time. Among all the data treatment models Map Reduce is mostly adopted. Due to the processing of data design in this method, once work is finished, the results are returned. The growth of data is not always a preferred batch operating environment: large numbers of applications can benefit from early results. The online aggregation method used in connection databases was first discussed. In this work analysis we discuss techniques that can allow early results to be estimated. We are proposing some changes to the online system for Map Reduce. We show that the properties required for accurate evaluation of the effects of our proposed device design changes. The reduction of data bias algorithm and sampling at the block levels are provided. Consequently, with a series of particular applications and sets of data, the implementation of the proposed design of the system is defined and evaluated.

show abstract

Section: Uniform Random Samplingmentioning

confidence: 99%