FAI: Advancing Deep Learning Towards Spatial Fairness

空间公平的目标是减少与数据样本的位置或地理区域有显著联系的偏差。如果听之任之，这种偏见可能会导致或加剧资源分配不公、社会分裂、空间差距以及复原力或可持续性方面的弱点。人工智能在农业监测和灾害管理等各种现实问题中的使用迫切需要空间公平。农产品，包括作物地图和种植面积估计，被用来为重要的决策提供信息，如发放补贴和提供农业保险。空间偏差造成的不准确和不平等对这些决定产生了不利影响。同样，有效和公平地绘制洪水或火灾等自然灾害的地图，对于为救生行动提供信息并量化公共基础设施的损害和风险至关重要，这与保险估计有关。机器学习，特别是深度学习，已经被广泛地应用于空间数据集，并取得了良好的结果。然而，由于数据分布、数据量和数据质量的变化，机器学习的直接应用在保持空间公平性方面的成功有限。该项目的目标是开发新一代学习框架，以明确维护空间公平。结果和代码将免费提供，并整合到现有的地理空间软件中。这些方法还将进行测试，以便纳入现有的实际系统(作物和水监测)。该项目旨在通过四个创新将深度学习方法推向空间公平。首先，将研究新的空间公平性统计公式，以应对连续空间域带来的独特挑战，特别是由于空间划分和创建位置组的各种方法，以及统计结论对空间划分的变化敏感的事实。其次，将开发新的网络体系结构，通过缓解由于数据分布在空间上的移动而导致的不同位置之间的冲突来提高空间公平性。第三，将使用新的公平驱动的对抗性学习策略来指导训练收敛到能够保持较高整体解质量的参数，同时最大化位置之间的空间公平性。最后，将提出一种知识增强型方法，该方法整合一般物理关系以缓解数据不平等引起的空间偏差，并模拟底层物理过程中的相关变量和参数以增强基于知识的空间公平性的可解释性。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The goal of spatial fairness is to reduce biases that have significant linkage to the locations or geographical areas of data samples. Such biases, if left unattended, can cause or exacerbate unfair distribution of resources, social division, spatial disparity, and weaknesses in resilience or sustainability. Spatial fairness is urgently needed for the use of artificial intelligence in a large variety of real-world problems such as agricultural monitoring and disaster management. Agricultural products, including crop maps and acreage estimates, are used to inform important decisions such as the distribution of subsidies and providing farm insurance. Inaccuracies and inequities produced by spatial biases adversely affect these decisions. Similarly, effective and fair mapping of natural disasters such as floods or fires is critical to inform live-saving actions and quantify damages and risks to public infrastructures, which is related to insurance estimation. Machine learning, in particular deep learning, has been widely adopted for spatial datasets with promising results. However, straightforward applications of machine learning have found limited success in preserving spatial fairness due to the variation of data distribution, data quantity, and data quality. The goal of this project is to develop a new generation of learning frameworks to explicitly preserve spatial fairness. The results and code will be made freely available and integrated into existing geospatial software. The methods will also be tested for incorporation in existing real systems (crop and water monitoring). This project aims to advance deep learning methods toward spatial fairness via four innovations. First, new statistical formulations of spatial fairness will be investigated to address unique challenges brought by the continuous spatial domain, particularly due to a variety of ways to partition the space and create location-groups for fairness evaluation, and the fact that statistical conclusions are sensitive to changes in space-partitionings. Second, new network architectures will be developed to improve the spatial fairness by mitigating the conflicts amongst different locations due to the shift of data distribution over space. Third, new fairness-driven adversarial learning strategies will be used to guide the training to converge to parameters that can maintain a high overall solution quality while maximizing spatial fairness across locations. Finally, a knowledge-enhanced approach will be proposed, which integrates general physical relationships to mitigate data-inequality incurred spatial biases, and simulates relevant variables and parameters in underlying physical processes to enhance knowledge-based interpretability of spatial fairness.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.