Fig. 3From: An exploration strategy improves the diversity of de novo ligands using deep reinforcement learning: a case for the adenosine A2A receptorMolecule generation with the assistance of the exploration strategy during the training process. For each step of token selection, a random variable was generated between 0 and 1. If the value is larger than a pre-set threshold (exploring rate, ε), the probability distribution is determined by the current generator (exploitation network, Gθ). Otherwise, it was determined by the exploration network (Gφ)Back to article page