Galaxy Zoo DECaLS: Detailed Visual Morphology Measurements from Volunteers and Deep Learning for 314,000 Galaxies (2102.08414v2)

Abstract: We present Galaxy Zoo DECaLS: detailed visual morphological classifications for Dark Energy Camera Legacy Survey images of galaxies within the SDSS DR8 footprint. Deeper DECaLS images (r=23.6 vs. r=22.2 from SDSS) reveal spiral arms, weak bars, and tidal features not previously visible in SDSS imaging. To best exploit the greater depth of DECaLS images, volunteers select from a new set of answers designed to improve our sensitivity to mergers and bars. Galaxy Zoo volunteers provide 7.5 million individual classifications over 314,000 galaxies. 140,000 galaxies receive at least 30 classifications, sufficient to accurately measure detailed morphology like bars, and the remainder receive approximately 5. All classifications are used to train an ensemble of Bayesian convolutional neural networks (a state-of-the-art deep learning method) to predict posteriors for the detailed morphology of all 314,000 galaxies. When measured against confident volunteer classifications, the networks are approximately 99% accurate on every question. Morphology is a fundamental feature of every galaxy; our human and machine classifications are an accurate and detailed resource for understanding how galaxies evolve.

• The paper introduces a catalog of 311K galaxies by integrating deep DECaLS imaging with volunteer classifications and Bayesian convolutional neural networks.
• The paper refines decision trees to capture subtle features like weak bars and merger disturbances, enhancing the accuracy of morphological analysis.
• The paper advances galaxy evolution studies by providing precise data that links detailed structures to star formation rates and quenching processes.

Overview of Galaxy Zoo DECaLS Data Release

This paper presents Galaxy Zoo DECaLS, a comprehensive catalog of detailed morphological classifications for 311,000 galaxies, utilizing the deep imaging capability provided by the Dark Energy Camera Legacy Survey (DECaLS). It highlights an innovative approach that integrates citizen science and advanced machine learning techniques to examine galaxy images, thereby offering a reliable dataset for understanding galaxy morphology and evolution.

Key Contributions

1. Deep Imaging Utilization: DECaLS provides deeper $r$-band imaging compared to previous surveys (extending to $r=23.6$ versus $r=22.2$ in SDSS), revealing finer structures, such as spiral arms and tidal features, which were not previously visible. This depth enhances the capability to paper subtle morphological features like bars and mergers.
2. Volunteer Classification Efforts: Leveraging the collective efforts of approximately three million responses from Galaxy Zoo volunteers, the paper incorporates an ensemble of Bayesian convolutional neural networks to extrapolate volunteer classifications to the entire sample of galaxies.
3. Improved Decision Trees: The paper introduces an enhanced decision tree with more nuanced responses for bars and mergers, distinguishing between strong and weak bars and differentiating types of disturbances due to mergers. This refinement aids in capturing subtle features influenced by galaxy dynamics and interactions.
4. Automated Classification Approach: By employing state-of-the-art Bayesian deep learning, the paper successfully predicts morphological features across grossly sampled galaxies beyond manual classification limits. The approach speculates on what volunteers would likely have identified from a small subset of volunteer-classified galaxies.
5. Implications on Galaxy Evolution Studies: Morphology serves as a key driver of galaxy evolution. The detailed catalog enables empirical investigations into the influence of structures like bars on star formation rates and the consequences of mergers on galactic assembly. Moreover, it facilitates exploration into auxiliary galaxy characteristics such as quenching processes.

Future Directions

The integration of volunteer-based classifications with automated methods in Galaxy Zoo presents an opportunity for scaling such combinations to upcoming massive surveys. Future focus could be oriented towards refining automated methods for improved accuracy on ambiguous galaxies and enhancing active learning strategies to optimize volunteer input. Continuous evaluation of biases and systematic errors remains crucial, especially with the rapidly evolving machine learning technologies applicable to astrophysics. These advancements promise a significant contribution to understanding large-scale cosmic phenomena through database enhancement and methodological innovations.