Noa
Yehezkel Lubin

Identifying Exoplanets Using Deep Learning Methods

Bar-Ilan University

Noa Lubin

Noa
Yehezkel Lubin

Identifying Exoplanets Using Deep Learning Methods

Bar-Ilan University

Noa Lubin

Bio

Noa is a Natural Language Processing Researcher at the BIU Computer Science Department.

Bio

Noa is a Natural Language Processing Researcher at the BIU Computer Science Department.

Abstract

TESS, Transiting Exoplanet Survey Satellite, is a critical mission to increase our understanding of earth-like planets outside our solar system. TESS will survey 200,000 of the brightest stars near us to search for exoplanets; i.e. planets outside our solar system. Given that TESS is surveying a different set of stars every 27 days, manual classification of threshold crossing events (TCEs) is very challenging.

 

Thus, we need machine classification systems able to automatically determine whether a TCE is a planet candidate (PC) or not. We have learned from our preliminary automatic classification study of Kepler’s TCEs that the features extracted by the pipeline might not be suitable for automatic classification. Therefore, in this work we focus on deep learning techniques that do both feature extraction and classification simultaneously.

Abstract

TESS, Transiting Exoplanet Survey Satellite, is a critical mission to increase our understanding of earth-like planets outside our solar system. TESS will survey 200,000 of the brightest stars near us to search for exoplanets; i.e. planets outside our solar system. Given that TESS is surveying a different set of stars every 27 days, manual classification of threshold crossing events (TCEs) is very challenging.


Thus, we need machine classification systems able to automatically determine whether a TCE is a planet candidate (PC) or not. We have learned from our preliminary automatic classification study of Kepler’s TCEs that the features extracted by the pipeline might not be suitable for automatic classification. Therefore, in this work we focus on deep learning techniques that do both feature extraction and classification simultaneously.

Discussion Points

  • First, how to decide whether a labeled data is a must? 
  • Different types of labeling challenges we’ve dealt with as data scientists (partial labels, noisy labels, etc.)
  • Academic approaches that discuss possible solutions to these problems
  • Practical solutions we eventually implemented 
  • Interesting case studies and results

Discussion Points

  • First, how to decide whether a labeled data is a must? 
  • Different types of labeling challenges we’ve dealt with as data scientists (partial labels, noisy labels, etc.)
  • Academic approaches that discuss possible solutions to these problems
  • Practical solutions we eventually implemented 
  • Interesting case studies and results

Planned Agenda

Planned Agenda