No abstract

The Kepler satellite has provided photometric timeseries data of unprecedented length, duty cycle and precision. To fully analyse these data for the tens of thousands of stars observed by Kepler, automated methods are a prerequisite. Here we present an automated procedure to determine the period spacing of gravity modes in redgiant stars ascending the red-giant branch. The gravity modes reside in a cavity in the deep interior of the stars and provide information on the conditions in the stellar core. However, for red giants the gravity modes are not directly observable on the surface, hence this method is based on the pressure-gravity mixed modes that present observable features in the Fourier power spectrum. The method presented here is based on the vertical alignment and symmetry of these mixed modes in a period echelle diagram. We find that we can obtain reliable results for both model frequencies and observed frequencies. Additionally, we carried out Monte Carlo tests to obtain realistic uncertainties on the period spacings with different set of oscillation modes (for the models) and uncertainties on the frequencies. Furthermore, this method has been used to improve mode detection and identification of the observed frequencies in an iterative manner.

Deep Learning for neuroimaging data is a promising but challenging direction. The high dimensionality of 3D MRI scans makes this endeavor compute and data-intensive. Most conventional 3D neuroimaging methods use 3D-CNN-based architectures with a large number of parameters and require more time and data to train. Recently, 2D-slice-based models have received increasing attention as they have fewer parameters and may require fewer samples to achieve comparable performance. In this paper, we propose a new architecture for BrainAGE prediction. The proposed architecture works by encoding each 2D slice in an MRI with a deep 2D-CNN model. Next, it combines the information from these 2D-slice encodings using set networks or permutation invariant layers. Experiments on the BrainAGE prediction problem, using the UK Biobank dataset, showed that the model with the permutation invariant layers trains faster and provides better predictions compared to other state-of-the-art approaches.

No abstract

A wealth of algorithms centered around (integer) linear programming have been proposed to compute equilibrium strategies in security games with discrete states and actions. However, in practice many domains possess continuous state and action spaces. In this paper, we consider a continuous space security game model with infinite-size action sets for players and present a novel deep learning based approach to extend the existing toolkit for solving security games. Specifically, we present (i) OptGradFP, a novel and general algorithm that searches for the optimal defender strategy in a parameterized continuous search space, and can also be used to learn policies over multiple game states simultaneously; (ii) OptGradFP-NN, a convolutional neural network based implementation of OptGradFP for continuous space security games. We demonstrate the potential to predict good defender strategies via experiments and analysis of OptGradFP and OptGradFP-NN on discrete and continuous game settings.