Machine learning models typically exert the same computational power on both easy and challenging examples. This is in stark contrast with humans, who use either fast (instinctive) or slow (analytical) thinking depending on the problem difficulty, a property called the dual-process theory of mind. We present FLOWGEN, a graph-generation model inspired by the dual-process theory of mind that generates large graphs incrementally. Depending on the difficulty of completing the graph at the current step, graph generation is routed to either a fast (weaker) or a slow (stronger) model. The fast and slow models have identical architectures, but vary in the number of parameters and consequently the strength. Experiments on real-world graphs show that ours can successfully generate graphs similar to those generated by a single large model in a fraction of time.

翻译：机器学习模型通常在简单和具有挑战性的例子上都使用相同的计算力。这与人类形成鲜明对比,人类使用快速(内向)或慢(分析)思维,视问题难度而定,这是称为双进程思维理论的属性。我们介绍了由双进程思维理论所启发的图形生成模型FlougGEN,该图形生成模型可逐渐生成大图。根据当前步骤完成图形的困难,图形生成可选择快速(粗)或慢(坚固)模型。快速和慢模式具有相同的结构,但参数数量和强度各不相同。在现实世界图上进行的实验显示,我们能够成功地生成与一个单一大模型在很短的时间里生成的相似的图形。