In this colloquium, I will briefly introduce our recent research using Deep Learning (DL) in astronomy, neural sciences, and condensed matter physics. In astronomy, our goal is to search for the Young Stellar Objects (YSOs) from the spectral energy distribution (SED) . We show that a YSO can be identified precisely even when using three SED bands only in the long wavelength regime, where the observational errors are much larger. In neural sciences, we identify the polarity of neuron cells from their optical image with a very high accuracy (>96%) even for complex neurons , making it possible to determine the direction of signal flows in the neural networks of a Drosophila brain. Finally, I will show how a DL can be used to solve the ground state properties of a strongly interacting many-body problem, using data obtained in the weak interacting regime . I hope this brief overview will demonstrate that how a DL could be also applied in fundamental research by providing deeper insights into our universe with multi-scales.
 Searching for Young Stellar Objects through SEDs by Machine Learning, Yi-Lung Chiu, Chi-Ting Ho, Daw-Wei Wang, and Shih-Ping Lai, Astronomy and Computing 36, 100470 (2021).
 High Accuracy Identification of Neuronal Polarity in the Insect Brain: a Node-Based Machine Learning Model, Chen-Zhi Su, Kuan-Ting Chou, Hsuan-Pei Huang, Chung-Chuan Lo, and Daw-Wei Wang, to be published in Neuroinformatics.
 Random Sampling Neural Network for Quantum Many-Body Problems, Chen-Yu Liu, Daw-Wei Wang, Phys. Rev. B 103, 205103 (2021).