Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Source Localizationby Inverse Diffusionand Convex Optimization
KTH, School of Engineering Sciences (SCI).
KTH, School of Engineering Sciences (SCI).
2018 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesisAlternative title
Källokaliseringmed inversdiffusion och konvex optimering (Swedish)
Abstract [sv]

Målet med rapporten har varit att att lokalisera källor till en emitterad substans med hjälp av en algoritm för inversdiffusion. Denna algoritm har gett lyckade resultat när den applicerats för biologisk detektering av celler och det har vidare föreslagits att körtiden skulle kunna reduceras avsevärt om algoritmen implementeras som en beräkningsgraf. Detta skulle automatisera beräkningar av gradienter och tillåta snabbare exekvering på en grafikprocessor. För algoritmens implementation användes TensorFlow, som primärt är ett programmeringsbibliotek inriktat mot maskininlärning. Datorgenererade testbilder av biologiska prover användes sedan för att utvärdera körresultatet med hjälp av mjukvara för bildanalys samt prestandamått. Jämförelser visar att implementationen i TensorFlow ger resultat som överensstämmer med den traditionella implementationen av samma algoritm. Sett ur ett bredare perspektiv visar detta på möjligheten att använda beräkningsgrafer för att lösa storskaliga optimeringsproblem och mer specifikt inversproblem.

Abstract [en]

The purpose of this project was to implement an inverse diffusion algorithm to locate the sources of an emitted substance. This algorithm has yielded successful results when applied to biological cell detection, and it has been suggested that the run time could be greatly reduced if adaptions for a computation graph framework are made. This would automate calculations of gradients and allow for faster execution on a graphics processing unit.

The algorithm implementation was realized in TensorFlow, which is primarily a machine learning oriented programming library. Computer-generated biological test images were then used to evaluate the performance using regular image analysis software and accuracy metrics.

Comparisons reveal that the TensorFlow implementation of the algorithm can match the accuracy metrics of traditional implementations of the same algorithm. Viewed in a broader scope this serves as an example to highlight the possibility of using computation graph frameworks to solve large scale optimization problems, and more specifically inverse problems.

 

Place, publisher, year, edition, pages
2018. , p. 11
Series
TRITA-SCI-GRU ; 2018-093
National Category
Engineering and Technology
Identifiers
URN: urn:nbn:se:kth:diva-230738OAI: oai:DiVA.org:kth-230738DiVA, id: diva2:1219083
Supervisors
Examiners
Available from: 2018-06-15 Created: 2018-06-15 Last updated: 2018-06-15Bibliographically approved

Open Access in DiVA

fulltext(623 kB)26 downloads
File information
File name FULLTEXT01.pdfFile size 623 kBChecksum SHA-512
f869b3a4348ed51ecb533159acc5e96c616adb4d61ab1f174a102eb669cbb2e8ffbecce5154022c5e27284e14d813c8339e5a3ef8612bf094a5af3f3e90696a8
Type fulltextMimetype application/pdf

By organisation
School of Engineering Sciences (SCI)
Engineering and Technology

Search outside of DiVA

GoogleGoogle Scholar
Total: 26 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 49 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf