1 AI与因果交织的城市复杂系统
1.1 复杂系统视角下的城市科学
核心问题：如何定量刻画城市(及内部要素)的空间组织、交互模式，及背后的动力学机制。
城市系统的空间组织模式与交互
引入：Zipf’s Law:
- Common power laws for cities and spatial fractal structures
- Zipf’s Law for Cities: An Explanation
考虑空间交互（人口流动）后：
- The growth equation of cities
空间组织模式：Christaller 中心地理论（Central Place Theory）: Hex lattice and optimization; Hyperuniformity（超均匀分布）
- Central places in Southern Germany
- A Note on Central Place Theory and the Range of a Good
- Hyperuniform organization in human settlements
空间交互：
- collective mobility：gravity model / intervention opportunity / radiation model / container model
- The scales of human mobility
- A universal model for mobility and migration patterns
- individual mobility: exploration and preferential returns model / visitation laws
- Modelling the scaling properties of human mobility
- The universal visitation law of human mobility
- 从交互到空间结构：polycentric transition / individual + collective movements
- Modeling the Polycentric Transition of Cities
- Emergence of urban growth patterns from human mobility behavior
城市系统中的标度律
核心问题:城市(及内部)的各种属性如何随人口规模系统性变化，及背后的动力学机制，解释城市集聚过程、规模效应的起源。
\[Y_t=Y_0N_t^{\beta}\]
- Allometry: The Study of Biological Scaling
- Growth, innovation, scaling, and the pace of life in cities
- Cities as Organisms: Allometric Scaling of Urban Road Networks
- The Origins of Scaling in Cities
- Simple spatial scaling rules behind complex cities
- Urban scaling laws arise from within-city inequalities
- Mathematical models to explain the origin of urban scaling laws
一些争议，城市边界的界定：
- Scaling: Lost in the Smog
- Reply to Huth et al.: Cities are defined by their spatially aggregated socioeconomic networks
空间网络，空间认知
- Neuronal vector coding in spatial cognition
- Spatial networks
- Networks and Cities: An Information Perspective
- Entropy of city street networks linked to future spatial navigation ability
- A new computational model for human navigation
一些新的观测结果/理论构建：
- Tracking job and housing dynamics with smartcard data
- Evidence for a conserved quantity in human mobility
对统一性框架建立的尝试：
- Urban growth and the emergent statistics of cities
其他可参考的书籍：
- Bettencourt, L. M. (2021). Introduction to urban science: evidence and theory of cities as complex svstems. MIT press.
- Barthelemy, M.(2019). The statistical physics of cities. Nature Reviews Physics, 1(6), 406-415.
- Barthelemy, M.(2016). The structure and dynamics of cities. Cambridge University Press.
1.2 城市科学中的AI
当下AI的大背景：On the Opportunities and Risks of Foundation Models
AI与地理学的渊源：
- 联结主义时期: Couclelis(1986): Artificial Intelligence in Geography: Conjecures on the Shape of Things to Come & Openshaw(1997): Artificial Intelligence in Geography
- 计算神经网络: Fischer(1998): Computational neural networks: a new paradigm for spatial analysis
- 地理人工智能: ML/DL applied in GIScience for Geographic Knowledge Discovery, Handbook of Geospatial Artificial Intelligence
spatial is special: 模式刻画/地统计/预测推测
空间显式模式：GeoAI: spatially explicit artificial intelligence techniques for geographic knowledge discovery and beyond
- 归纳偏置（inductive bias）：对学习的问题做的假设
- 空间智能分析：空间形式化 / 弱统计先验 / 侧重预测和机理解释
- 城市计算：Urban Computing: Concepts, Methodologies, and Applications
- 社会感知：Social sensing: A new approach to understanding our socioeconomic environments
GeoAI的一些问题：
- explanability 可解释性：Extracting spatial effects from machine learning model using local interpretation method: An example of SHAP and XGBoost
- weak replicability 弱可复现性（<- 空间异质性，统计分布随时空变化）: Replication across space and time must be weak inthe social and environmental sciences
- 预测误差的空间偏差
1.3 城市科学中的因果性
- 基于相关：如果没有变化，会观测到什么
- 基于因果：如果发生变化，会观测到什么
准实验方法，基于观察实验设计实践：
基于观察数据的ML算法：
- 基于约束的方法：条件独立测试
- 基于函数模型：从因果机制出发
- 基于评分-搜索：将因果发现视作组合优化问题
- 混合型方法
GIS领域的因果推断：Causal Inference in Spatial Statistics