Högskolan i Skövde

his.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • apa-cv
  • 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
Investigate the impact of feature selection on the performance of deep feedforward neural network
University of Skövde, School of Bioscience.
2023 (English)Independent thesis Advanced level (degree of Master (Two Years)), 30 credits / 45 HE creditsStudent thesis
Abstract [en]

Artificial Neural Networks are algorithms that are inspired by neural signaling in the human brain. Deep learning is a subset of machine learning methods that utilizes artificial neural networks and feed-forward neural networks are a common artificial neural network that is applied when using deep learning methods. Feature selection works by selecting features that are relevant for classification and thus, allows for easier interpretation of the model as well as reduction of overfitting. The aim of this study was to investigate the impact of feature selection on the performance of the model. This was achieved by performing a binary classification analysis, where the overall five year survivial of patients that suffered from breast cancer was predicted. A deep feed forward neural network model was trained to differentiate between “alive” and “deceased” patients. As a part of the procedure, different feature selection methods were used to observe the effect that they had on the model  performance. According to the results, using feature selection with mutual information, variance filter and PCA had an impact on the deep FFNN model performance, as compared to the correlation based method, which was significantly different in the model performance. In conclusion, feature selection with mutual information, variance filter and PCA lead to a better model performance, both in the discriminative power as well as the overall ability to make correct predictions as compared to correlation-based feature selection, which was proven to provide a poor model performance.

Place, publisher, year, edition, pages
2023. , p. 30
National Category
Bioinformatics and Computational Biology
Identifiers
URN: urn:nbn:se:his:diva-22880OAI: oai:DiVA.org:his-22880DiVA, id: diva2:1777289
Subject / course
Systems Biology
Educational program
Infection Biology - Master’s Programme 120 ECTS
Supervisors
Examiners
Available from: 2023-06-29 Created: 2023-06-29 Last updated: 2025-09-29Bibliographically approved

Open Access in DiVA

No full text in DiVA

By organisation
School of Bioscience
Bioinformatics and Computational Biology

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 472 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • apa-cv
  • 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