@inproceedings{behl2020autosimulate,
author = {Behl, Harkirat Singh and Baydin, Atılım Güneş and Gal, Ran and Torr, Philip H.S. and Vineet, Vibhav},
title = {AutoSimulate: (Quickly) Learning Synthetic Data Generation},
booktitle = {16th European Conference Computer Vision (ECCV 2020)},
year = {2020},
month = {August},
abstract = {Simulation is increasingly being used for generating large labelled datasets in many machine learning problems. Recent methods have focused on adjusting simulator parameters with the goal of maximising accuracy on a validation task, usually relying on REINFORCElike gradient estimators. However these approaches are very expensive as they treat the entire data generation, model training, and validation pipeline as a black-box and require multiple costly objective evaluations at each iteration. We propose an efficient alternative for optimal synthetic data generation, based on a novel differentiable approximation of the objective. This allows us to optimize the simulator, which may be non-differentiable, requiring only one objective evaluation at each iteration with a little overhead. We demonstrate on a state-of-the-art photorealistic renderer that the proposed method finds the optimal data distribution faster (up to 50×), with significantly reduced training data generation and better accuracy on real-world test datasets than previous methods.},
url = {http://approjects.co.za/?big=en-us/research/publication/autosimulate-quickly-learning-synthetic-data-generation/},
}