− | 现实中,当我们要应用涌现的量化框架时,会出现微观状态的数据难以获取、粗粒化策略不好确定的问题,所以,我们需要找到从可观测数据中直接识别因果涌现的方法。
| + | 上述理论框架很好地定量刻画了各种涌现现象。但是在现实中,我们更需要解决的是从数据中辨识出因果涌现是否发生的问题。从数据中识别出复杂系统的因果涌现,一方面可以节约预测成本,增强预测效果,尤其是泛化效果;另一方面也可能诱导或者预防涌现的发生。在实际应用中,一些微观数据难以捕捉,我们往往只能收集到观测数据,并推断系统的真实动力学。因此,从可观测数据中识别系统中因果涌现的发生是一个关键的问题。 |
− | 识别出复杂系统中的因果涌现,一方面可以节约预测成本,增强预测效果,尤其是泛化效果;另一方面也可能诱导或者预防涌现的发生。在实际应用中,一些微观数据难以捕捉,我们往往只能收集到观测数据,并推断系统的真实动力学。因此,从可观测数据中识别系统中因果涌现的发生是一个关键的问题。近年来,基于神经网络的机器学习方法取得了突破性进展<ref>Vlachas P-R, Arampatzis G and Uhler C et al. Multiscale simulations of complex systems by learning their effective dynamics. Nat Mach Intell 2022; 4: 359–366.</ref><ref>Kemeth F-P, Bertalan T and Thiem T et al. Learning emergent partial differential equations in a learned emergent space. Nat Commun 2022; 13: 3318.</ref><ref>Floryan D and Graham M-D. Data-driven discovery of intrinsic dynamics. Nat Mach Intell 2022; 4: 1113–1120.</ref><ref>Cai L and Ji S. A multi-scale approach for graph link prediction. Proceedings of the AAAI Conference on Artificial Intelligence, New York, 20-27 February 2020.</ref><ref>Chen Z, Li S and Yang B et al. Multi-scale spatial temporal graph convolutional network for skeleton-based action recognition. Proceedings of the AAAI Conference on Artificial Intelligence, New York, 22 February - 1 March 2022.</ref>,借助此方法,以数据驱动的方式自主发现复杂系统的因果关系甚至动力学成为可能。此外,机器学习和神经网络还可以帮助我们找到更好的粗粒化策略、从数据中发现宏观层面的因果关系。
| + | 近年来,基于神经网络的机器学习方法取得了突破性进展<ref>Vlachas P-R, Arampatzis G and Uhler C et al. Multiscale simulations of complex systems by learning their effective dynamics. Nat Mach Intell 2022; 4: 359–366.</ref><ref>Kemeth F-P, Bertalan T and Thiem T et al. Learning emergent partial differential equations in a learned emergent space. Nat Commun 2022; 13: 3318.</ref><ref>Floryan D and Graham M-D. Data-driven discovery of intrinsic dynamics. Nat Mach Intell 2022; 4: 1113–1120.</ref><ref>Cai L and Ji S. A multi-scale approach for graph link prediction. Proceedings of the AAAI Conference on Artificial Intelligence, New York, 20-27 February 2020.</ref><ref>Chen Z, Li S and Yang B et al. Multi-scale spatial temporal graph convolutional network for skeleton-based action recognition. Proceedings of the AAAI Conference on Artificial Intelligence, New York, 22 February - 1 March 2022.</ref>,借助此方法,以数据驱动的方式自主发现复杂系统的因果关系、因果机制,以至动力学都成为可能。此外,机器学习和神经网络还可以帮助我们找到更好的粗粒化策略、从数据中发现宏观层面的因果关系。 |