kth.sePublications
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
Regression with Bayesian Confidence Propagating Neural Networks
KTH, School of Electrical Engineering and Computer Science (EECS).
2023 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesisAlternative title
Regression med Bayesian Confidence Propagating Neural Networks (Swedish)
Abstract [en]

Bayesian Confidence Propagating Neural Networks (BCPNNs) are biologically inspired artificial neural networks. These networks have been modeled to account for brain-like aspects such as modular architecture, divisive normalization, sparse connectivity, and Hebbian learning. Recent research suggests that these networks have achieved high performance in classification-based tasks with additions such as structural plasticity. There is, however, scarce research on modeling BCPNN in regression-based tasks. In modeling these tasks, we aim to mimic continuous codes in the brain. Such codes have been shown to describe functions such as brain-motor activity. In this thesis, we conduct a systematic analysis to compare a baseline model with extensions of the BCPNN architecture that include two encoding techniques: Interval encoding and Gaussian Mixture Model (GMM) encoding. We also extend the model with a Ridge regressor. Our systematic analysis shows that this BCPNN model adapted with GMM encoding outperforms others. Moreover, the encodings also preserve sparse activity.

Abstract [sv]

Bayesian Confidence Propagating Neural Networks (BCPNN) är biologiskt inspirerade artificiella neurala nätverk. Dessa nätverk har modellerats för att ta hänsyn till hjärnliknande aspekter som modulär arkitektur, splittande normalisering, sparsam anslutning och hebbisk inlärning. Ny forskning tyder på att dessa nätverk har uppnått hög prestanda i klassificeringsbaserade uppgifter med tillägg som strukturell plasticitet. Det finns dock knappt forskning om modellering av BCPNN i regressionsbaserade uppgifter. Vid modellering av dessa uppgifter strävar vi efter att efterlikna kontinuerliga koder i hjärnan. Sådana koder har visat sig beskriva funktioner som hjärnmotorisk aktivitet. I detta examensarbete genomför vi en systematisk analys för att jämföra en baslinjemodell med utökningar av BCPNN-arkitekturen som inkluderar två kodningstekniker: Intervallkodning och Gaussisk blandningsmodell (GMM). Vi utökar även modellen med en Ridge-regressor. Vår systematiska analys visar att denna BCPNN-modell anpassad med GMM-kodning generellt sett överträffar andra. Dessutom bevarar kodningarna också sparsam aktivitet.

Place, publisher, year, edition, pages
2023. , p. 55
Series
TRITA-EECS-EX ; 2023:22
Keywords [en]
Machine Learning, Neural Networks, Brain-like computing, Bayesian Confidence Propagating Neural Networks
Keywords [sv]
Maskininlärning, neurala nätverk, hjärnliknande datorer, Bayesian Förtroendespridande neurala nätverk
National Category
Computer and Information Sciences
Identifiers
URN: urn:nbn:se:kth:diva-324209OAI: oai:DiVA.org:kth-324209DiVA, id: diva2:1738778
Supervisors
Examiners
Available from: 2023-02-25 Created: 2023-02-22 Last updated: 2023-02-25Bibliographically approved

Open Access in DiVA

fulltext(1217 kB)351 downloads
File information
File name FULLTEXT01.pdfFile size 1217 kBChecksum SHA-512
f256390ea81c347980d2ab4a0655bf0e0d4061d4b8081da4b2f0dfdf500addcd876ecbeaf8925db3cdde50729cb33725b3d0f5e6765424602beb94fdb19b857c
Type fulltextMimetype application/pdf

By organisation
School of Electrical Engineering and Computer Science (EECS)
Computer and Information Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 351 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: 596 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