Fig. 7From: Molecular optimization by capturing chemist’s intuition using deep neural networksData distribution over pairwise property changes on the training set where the circle size corresponds to counts. In a, b each tick x in horizontal axis represents \((x-0.1, x+0.1]\), e.g. 0 represents \((-\,0.1, 0.1]\). For ease of presentation, \(x<-1.6\) and \(x>1.6\) are not shown hereBack to article page