/GAN-abstraction

Primary LanguagePapyrus

GAN-abstraction

Abstraction of Markov Population Dynamics via Generative Adversarial Nets

Luca Bortolussi, Francesca Cairoli, Ginevra Carbone.

The code containing the experiments and the results presented at the TACAS 21 conference are located in the TACAS_21 folder. The folder is organized as follows:

  • Stochpy/psc_models folder contains the CRN model expressed, as a set of reactions, in the Stochpy format.
  • Dataset_Generation folder that contains the scripts to generate the datasets through SSA simulation (based on Stochpy library):
  • WGAN_for_Trajectories folder that contains the implementation of teh c-WCGAN with either fixed parameters (class WGAN_MA_fixed_param) or with an arbitrary number of varying parameters (class WGAN_MA).

The datasets used are shared at the following link: https://www.dropbox.com/sh/37dm23flg6k803e/AAA-Z314lCDTb1dA_IZb9RgEa?dl=0