 |
Under these backgrounds, high-performance computing systems also produce a large volume of data.
However, it is practically infeasible to send all raw data to the remote site for processing.
We should need a new approach to manage data efficiently and reliably data management. We have four stages for data management. The first is to collect data.
The second is to compress data. The third is to save data. Last is to reconstruct data from data compression.
We compress datasets using lossy compression algorithms based on three transformations, namely, discrete cosine transform (DCT).
Consequently, the objectives of my presentation are to evaluate the effect of lossy compression and present a model for the prediction of compression ratio.
From now, I will talk in detail about the impact of lossy compression-based DCT-transform in point data compression.
These pictures show the example of DCT transformation.
As previously mentioned, we use DCT signal transformation approaches for compressing data.
As you know, this is a standard signal processing algorithm. So I do not explain the detailed algorithms
But, I will explain how to compress using this in the next slide.
|
IEEE/ICACT20220021 Slide.05
[Big Slide]