Deep Reinforcement Learning for Data-Driven Adaptive Scanning in Ptychography

TL;DR

This paper introduces a deep reinforcement learning-based adaptive scanning method for ptychography that reduces the required dose while maintaining high-resolution reconstruction, outperforming traditional methods.

Contribution

It presents a novel RL-trained deep learning approach for adaptive scanning in ptychography, improving resolution at lower doses compared to conventional techniques.

Findings

Adaptive scanning achieves higher resolution at lower doses

Reinforcement learning effectively guides the scanning process

Method outperforms conventional ptychography in experiments

Abstract

We present a method that lowers the dose required for a ptychographic reconstruction by adaptively scanning the specimen, thereby providing the required spatial information redundancy in the regions of highest importance. The proposed method is built upon a deep learning model that is trained by reinforcement learning (RL), using prior knowledge of the specimen structure from training data sets. We show that equivalent low-dose experiments using adaptive scanning outperform conventional ptychography experiments in terms of reconstruction resolution.