InterOpApr2022KDIG < IVOA

---++ Knowledge Discovery 1

Time: [[https://www.timeanddate.com/worldclock/fixedtime.html?msg=KD&iso=20220428T2200][Thursday Apr 28 22:00 UTC]]

| <p> _Raffaele D'Abrusco_ </p> | <p> *Introduction* </p> | 5' | |
| <p>Rafael Martinez Galarza</p> | | 15' | |
| _Petr Skoda_ | <p>The Role of VO Technology in Astronomical Machine<br />Learning</p> | 15' | The VO infrastructure has already increased the efficiency of<br />traditional astronomical data analysis in many ways. We try to identify<br />the points where the IVOA standards (current or after some modifications)<br />can make easier the preparation of machine learning experiments or even<br />introduce new KDD methodology. |
| _Yihan Tao_ | <p>Classification of Galaxy Spectra based on Convolutional Neural<br />Network</p> | 15' | It is important to classify galaxies by their spectral data<br />in astronomy. The widely used spectral classification method for galaxy<br />is the BPT diagram, which classifies emission line galaxies by flux<br />ratios of the Balmer and forbidden lines. A convolutional neural network<br />(CNN) is a deep learning algorithm which achieves outstanding<br />performance on feature extraction and classification tasks. We build a<br />one-dimensional CNN model to classify galaxy spectra into star-forming,<br />composite, AGN, and normal galaxies. We preprocessed the galaxy spectra<br />selected from SDSS DR16 with spectral interval constraint and flux<br />standardization and conducted experiments with the preprocessed data.<br />The dataset labels are derived from the intersection of the BPT<br />classification given by MPA-JHU and Portsmouth catalog. The results<br />showed that the classification accuracy of our network was over 92%<br />without relying on redshifts. |

Moderator: [[Raffaele D'Abrusco][Raffaele D'Abrusco]], Notetaker: [[IVOA.TBD][TBD]], [[https://yopad.eu/p/IVOA_Nov3_KD][Etherpad link]]
This topic: IVOA > WebHome > WebPreferences > ProgramPrepVirtualApr2022 > InterOpApr2022 > InterOpApr2022KDIG
Topic revision: r4 - 2022-04-12 - RaffaeleDAbrusco