國立虎尾科技大學 |

Compact and Progressive Encodings for Task-Adaptive Queries of Scientific Data.

Record Type:	Language materials, manuscript : Monograph/item
Title/Author:	Compact and Progressive Encodings for Task-Adaptive Queries of Scientific Data./
Author:	Hoang, Duong.
Description:	1 online resource (215 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 85-04, Section: A.
Contained By:	Dissertations Abstracts International85-04A.
Subject:	Computer science. -
Online resource:	click for full text (PQDT)
ISBN:	9798380590150

Compact and Progressive Encodings for Task-Adaptive Queries of Scientific Data.
Hoang, Duong.

Compact and Progressive Encodings for Task-Adaptive Queries of Scientific Data. - 1 online resource (215 pages)

Source: Dissertations Abstracts International, Volume: 85-04, Section: A.

Thesis (Ph.D.)--The University of Utah, 2023.

Includes bibliographical references

As computational power outpaces bandwidth, handling multi-petabyte datasets, which are easier to produce than to store or access, becomes a challenge. Traditional solutions focus on data reduction through lossy compression or downsampling, but many overlook considerations like the scientific tasks to be performed on the data or the computational resources required. Additionally, there's a lack of a common framework to gauge these techniques' efficacy and costs. This dissertation introduces such a framework and presents novel data reduction techniques offering more holistic considerations and compelling trade-offs.A novel hierarchical data model that unifies spatial resolution and numerical precision is developed. Most current reduction techniques focus on either resolution or precision, but this new model allows for approximations anywhere in the 2D space of resolution and precision, offering refined approximation in either axis without necessarily refining the other.A novel progressive streaming framework is proposed to study trade-offs between reducing resolution and precision during common analysis tasks. Using the wavelet transform, both resolution and precision-based reduction techniques can be modeled as different orderings of wavelet coefficient bits. This enables the first-ever demonstration that reducing both resolution and precision is significantly more effective than reducing either alone. It further allows the optimization of bit streams for specific analysis tasks, offering insights about how different tasks fetch data.Lastly, the new data model is adapted to encode unstructured particle data, a common form of data representation in scientific simulation or imaging. Borrowing techniques from the structured case, it's shown that the odd-even decomposition can significantly reduce reconstruction error while maintaining a constant memory footprint. This results in novel particle hierarchies and optimal traversal heuristics that outperform traditional approaches in progressive decoding quality-cost trade-offs.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024

Mode of access: World Wide Web

ISBN: 9798380590150Subjects--Topical Terms:

573171
Computer science.
Subjects--Index Terms:

Computational resourcesIndex Terms--Genre/Form:

554714
Electronic books.

Compact and Progressive Encodings for Task-Adaptive Queries of Scientific Data.
LDR:03414ntm a22003857 4500 001 1147275
005 20240909100744.5
006 m o d
007 cr bn ---uuuuu
008 250605s2023 xx obm 000 0 eng d
020 $a 9798380590150
035 $a (MiAaPQ)AAI30570757
035 $a AAI30570757
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Hoang, Duong. $3 1472965
245 1 0 $a Compact and Progressive Encodings for Task-Adaptive Queries of Scientific Data.
264 0 $c 2023
300 $a 1 online resource (215 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 85-04, Section: A.
500 $a Advisor: Pascucci, Valerio.
502 $a Thesis (Ph.D.)--The University of Utah, 2023.
504 $a Includes bibliographical references
520 $a As computational power outpaces bandwidth, handling multi-petabyte datasets, which are easier to produce than to store or access, becomes a challenge. Traditional solutions focus on data reduction through lossy compression or downsampling, but many overlook considerations like the scientific tasks to be performed on the data or the computational resources required. Additionally, there's a lack of a common framework to gauge these techniques' efficacy and costs. This dissertation introduces such a framework and presents novel data reduction techniques offering more holistic considerations and compelling trade-offs.A novel hierarchical data model that unifies spatial resolution and numerical precision is developed. Most current reduction techniques focus on either resolution or precision, but this new model allows for approximations anywhere in the 2D space of resolution and precision, offering refined approximation in either axis without necessarily refining the other.A novel progressive streaming framework is proposed to study trade-offs between reducing resolution and precision during common analysis tasks. Using the wavelet transform, both resolution and precision-based reduction techniques can be modeled as different orderings of wavelet coefficient bits. This enables the first-ever demonstration that reducing both resolution and precision is significantly more effective than reducing either alone. It further allows the optimization of bit streams for specific analysis tasks, offering insights about how different tasks fetch data.Lastly, the new data model is adapted to encode unstructured particle data, a common form of data representation in scientific simulation or imaging. Borrowing techniques from the structured case, it's shown that the odd-even decomposition can significantly reduce reconstruction error while maintaining a constant memory footprint. This results in novel particle hierarchies and optimal traversal heuristics that outperform traditional approaches in progressive decoding quality-cost trade-offs.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2024
538 $a Mode of access: World Wide Web
650 4 $a Computer science. $3 573171
650 4 $a Computer engineering. $3 569006
650 4 $a Information science. $3 561178
653 $a Computational resources
653 $a Spatial resolution
653 $a Scientific data
653 $a Borrowing techniques
655 7 $a Electronic books. $2 local $3 554714
690 $a 0984
690 $a 0464
690 $a 0723
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a The University of Utah. $b School of Computing. $3 1188323
773 0 $t Dissertations Abstracts International $g 85-04A.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30570757 $z click for full text (PQDT)