In this project, I tried to reproduce the pseudo-code of a Guided Evolution Strategies using Historical Gradients form this article https://www.ijcai.org/proceedings/2020/0205.pdf.
The implementation was done on Tensorflow and Python. The algorithms were tested on the Nevergrad function https://facebookresearch.github.io/nevergrad/ and on the OpenGym AI control suite https://gym.openai.com/docs/.
The first document corresponds to the report of this project and the second document is a Jupyter Notebook displaying the code of the algorithm and the results quoted in the report.