AutoML Research

PyGlove is a versatile library that not only simplifies the implementation of AutoML applications but is also powerful enough to facilitate complex AutoML/ML research. Its capability and flexibility have been demonstrated in several academic papers, including the following:

Papers with code:

• Evolving Reinforcement Learning algorithms, ICLR 2021 [code]

• PyGlove: Efficiently Exchanging ML Ideas as Code, 2022 [code]

Papers only:

• PyGlove: Symbolic Programming for Automated Machine Learning, NeurIPS 2020

• AutoHAS: Efficient Hyperparameter and Architecture Search, ICLR 2021 NAS workshop

• Towards the Co-design of Neural Networks and Accelerators, MLSys 2022

• Deepfusion: Lidar-camera Deep Fusion for Multi-modal 3D Object Detection, CVPR 2022

• ES-ENAS: Combining Evolution Strategies with Neural Architecture Search at No Extra Cost for Reinforcement Learning, CoRR 2021