Deep learning implementation for Dual-Readout calorimeter.

The dual-readout calorimeter consists of scintillating and Cerenkov fibers readout together. This design allows both electromagnetic and hadronic showers to be measured with high precision in a single detector. While it's under development for future colliders, deep learning implementations are studied to maximize the physics potential. Image based deep learning model analyzes pixelated energy deposit with convolutional neural networks. And raw energy deposit can be applied with a point cloud based deep learning method. Using these methods, jet reconstruction, particle identification, and fast simulation can be improved. We present demonstrations of jet variables regression, particle discrimination, and shower generator for the dual-readout calorimeter.