Point Cloud Generation using Transformer Encoders and Normalising Flows

TL;DR

This paper introduces a novel point cloud generation method combining normalising flows and transformer encoders, achieving state-of-the-art results and stable training for complex collider data simulation.

Contribution

It presents a new model integrating normalising flows with transformer encoders, trained adversarially, for improved point cloud data generation in particle physics.

Findings

Achieves state-of-the-art performance in collider data generation

Provides stable training process for complex point clouds

Outperforms previous models in accuracy and efficiency

Abstract

Data generation based on Machine Learning has become a major research topic in particle physics. This is due to the current Monte Carlo simulation approach being computationally challenging for future colliders, which will have a significantly higher luminosity. The generation of collider data is similar to point cloud generation, but arguably more difficult as there are complex correlations between the points which need to be modelled correctly. A refinement model consisting of normalising flows and transformer encoders is presented. The normalising flow output is corrected by a transformer encoder, which is adversarially trained against another transformer encoder discriminator/critic. The model reaches state-of-the-art performance while yielding a stable training.