Development of Knowledge Extraction Methodology from Trained Artificial Neural Networks
Promocijas darba kopsavilkums
Andrejs Bondarenko, Rīgas Tehniskā universitāte, Latvija
http://orcid.org/0000-0002-4103-6814
Artificial neural networks (ANN) are widely used in machine learning. They are powerful non-linear models that can be trained in a supervised, semi-supervised, and unsupervised manner. There is no single best machine learning classifier that can be used in all scenarios, but ANNs are frequently outperforming other classifiers. On the downside, it is hard to explain how classification decision is made within ANN. Artificial neural networks are essentially black-boxes. Lack of understanding of how such classifiers work severely limits their applicability. The Thesis is devoted to the development of approaches allowing to extract knowledge in the form of rules from trained ANN classifier.
Papildus informācija
| DOI | |
|---|---|
| Izdevuma tips | |
| Hipersaite | |
| Aizstāvēšanas datums | 18.06.2020. |
| ISBN (print) | 978-9934-22-011-1 |
| ISBN (pdf) | 978-9934-22-140-8 |
| Formāts | |
| Lappušu skaits | 44 |
| Publicēts tiešsaistē | |
| Izdevējs | RTU Izdevniecība |
| Publicēšanas valsts | Latvija |
| Valoda |
You must be logged in to post a review.
Atsauksmes
Vēl nav nevienas atsauksmes.