Abstract
Spatial variability is a prominent feature of various geographic phenomena such as climatic zones, USDA plant hardiness zones, and terrestrial habitat types (e.g., forest, grasslands, wetlands, and deserts). However, current deep learning methods follow a spatial-one-size-fits-all (OSFA) approach to train single deep neural network models that do not account for spatial variability. Quantification of spatial variability can be challenging due to the influence of many geophysical factors. In preliminary work, we proposed a spatial variability aware neural network (SVANN-I, formerly called SVANN) approach where weights are a function of location but the neural network architecture is location independent. In this work, we explore a more flexible SVANN-E approach where neural network architecture varies across geographic locations. In addition, we provide a taxonomy of SVANN types and a physics inspired interpretation model. Experiments with aerial imagery based wetland mapping show that SVANN-I outperforms OSFA and SVANN-E performs the best of all.
- [1] . 2017. SegNet: A deep convolutional encoder-decoder architecture for image segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence 39, 12 (2017), 2481–2495.Google ScholarCross Ref
- [2] . 2003. Spatial variability of soil carbon in forested and cultivated sites: Implications for change detection. Journal of Environmental Quality 32, 1 (2003), 278–286.Google Scholar
- [3] . 1979. Classification of Wetlands and Deepwater Habitats of the United States. Fish and Wildlife Service, U.S. Department of the Interior.Google ScholarCross Ref
- [4] . 1989. Approximations by superpositions of a sigmoidal function. Mathematics of Control, Signals and Systems 2 (1989), 183–192.Google ScholarCross Ref
- [5] . 1988. An exponential model used for optimal threshold selection on ROC curves. Medical Decision Making 8, 2 (1988), 120–131.Google ScholarCross Ref
- [6] . 2017. Machine learning for medical imaging. Radiographics 37, 2 (2017), 505–515.Google ScholarCross Ref
- [7] . 2020. National Wetlands Inventory: Surface Waters and Wetlands. Department of the Interior, Fish and Wildlife Service, Washington, DC. https://fwsprimary.wim.usgs.gov/wetlands/apps/wetlands-mapper/.Google Scholar
- [8] . 2003. Geographically Weighted Regression: The Analysis of Spatially Varying Relationships. John Wiley & Sons.Google Scholar
- [9] . 1996. Experiments with a new boosting algorithm. In Proceedings of the 13th International Conference on Machine Learning (ICML’96). 148–156. Google ScholarDigital Library
- [10] . 2016. Deep learning for visual understanding: A review. Neurocomputing 187 (2016), 27–48. Google ScholarDigital Library
- [11] . 2020. Towards spatial variability aware deep neural networks (SVANN): A summary of results. In Proceedings of the Workshop on Deep Learning for Spatiotemporal Data, Applications, and Systems (DeepSpatial’20).Google Scholar
- [12] . 2015. Mapping spatial heterogeneity in the tumor microenvironment: A new era for digital pathology. Laboratory Investigation 95, 4 (2015), 377–384.Google ScholarCross Ref
- [13] . 2018. TernausNet: U-Net with VGG11 encoder pre-trained on ImageNet for image segmentation. arXiv preprint arXiv:1801.05746 (2018).Google Scholar
- [14] . 2019. Spatial ensemble learning for heterogeneous geographic data with class ambiguity. ACM Transactions on Intelligent Systems and Technology 10, 4 (2019), 1–25. Google ScholarDigital Library
- [15] . 2010. Wetland Ecology: Principles and Conservation. Cambridge University Press.Google ScholarCross Ref
- [16] . 2018. Algorithms for semantic segmentation of multispectral remote sensing imagery using deep learning. ISPRS Journal of Photogrammetry and Remote Sensing 145 (2018), 60–77.Google ScholarCross Ref
- [17] . 2015. A semi-automated, multi-source data fusion update of a wetland inventory for East-Central Minnesota, USA. Wetlands 35, 2 (2015), 335–348.Google ScholarCross Ref
- [18] . 2012. ImageNet classification with deep convolutional neural networks. In Advances in Neural Information Processing Systems. 1097–1105. Google ScholarDigital Library
- [19] . 2003. Measures of diversity in classifier ensembles and their relationship with the ensemble accuracy. Machine Learning 51, 2 (2003), 181–207. Google ScholarDigital Library
- [20] . 2015. Deep learning. Nature 521, 7553 (2015), 436.Google ScholarCross Ref
- [21] . 2019. Land cover segmentation of aerial imagery using SegNet. In Earth Resources and Environmental Remote Sensing/GIS Applications X, Vol. 11156. International Society for Optics and Photonics, 111561J.Google Scholar
- [22] . 2018. Laws of geography. In Oxford Research Encyclopedia of Criminology and Criminal Justice. Oxford University Press, Oxford, UK.
DOI: 10.1093/acrefore/9780190264079.013.325Google Scholar - [23] . 2008. Introduction to Information Retrieval. Cambridge University Press. Google ScholarCross Ref
- [24] . 2018. Deep learning for healthcare: Review, opportunities and challenges. Briefings in Bioinformatics 19, 6 (2018), 1236–1246.Google ScholarCross Ref
- [25] . 1997. The National Resources Inventory: A long-term multi-resource monitoring programme. Environmental and Ecological Statistics 4, 3 (1997), 181–204.Google ScholarCross Ref
- [26] . 2012. USDA Plant Hardiness Zone Map. USDA. https://planthardiness.ars.usda.gov/PHZMWeb/.Google Scholar
- [27] . 2009. A survey on transfer learning. IEEE Transactions on Knowledge and Data Engineering 22, 10 (2009), 1345–1359. Google ScholarDigital Library
- [28] . 2017. YOLO9000: Better, faster, stronger. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 7263–7271.Google ScholarCross Ref
- [29] . 2015. U-Net: Convolutional networks for biomedical image segmentation. In Proceedings of the International Conference on Medical Image Computing and Computer-Assisted Intervention. 234–241.Google ScholarCross Ref
- [30] . 2002. Artificial Intelligence: A Modern Approach (2nd ed.). Prentice Hall.Google Scholar
- [31] . 2017. Training deep convolutional neural networks for land–cover classification of high-resolution imagery. IEEE Geoscience and Remote Sensing Letters 14, 4 (2017), 549–553.Google ScholarCross Ref
- [32] . 2012. Spatial big-data challenges intersecting mobility and cloud computing. In Proceedings of the 11th ACM International Workshop on Data Engineering for Wireless and Mobile Access. 1–6. Google ScholarDigital Library
- [33] . 2009. A new dawn for citizen science. Trends in Ecology & Evolution 24, 9 (2009), 467–471.Google ScholarCross Ref
- [34] . 2014. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556 (2014).Google Scholar
- [35] . 2018. An Overview of Rainfall-Runoff Model Types. U.S. Environmental Procection Agency, Washington, DC.Google Scholar
- [36] . 2015. Going deeper with convolutions. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 1–9.Google ScholarCross Ref
- [37] . 1990. Use of high-altitude aerial photography for inventorying forested wetlands in the United States. Forest Ecology and Management 33, 34 (1990), 593–604.Google ScholarCross Ref
- [38] . 2005. Causes and consequences of spatial heterogeneity in ecosystem function. In Ecosystem Function in Heterogeneous Landscapes. Springer, 9–30.Google ScholarCross Ref
- [39] . 1982. Simpson’s paradox in real life. American Statistician 36, 1 (1982), 46–48.Google Scholar
- [40] . 2020. A locally-constrained YOLO framework for detecting small and densely-distributed building footprints. International Journal of Geographical Information Science 34, 4 (2020), 777–801.Google ScholarCross Ref
- [41] . 2019. Revolutionizing tree management via intelligent spatial techniques. In Proceedings of the 27th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems. 71–74. Google ScholarDigital Library
- [42] . 2017. Deep learning in remote sensing: A comprehensive review and list of resources. IEEE Geoscience and Remote Sensing Magazine 5, 4 (2017), 8–36.Google ScholarCross Ref
Index Terms
- Spatial Variability Aware Deep Neural Networks (SVANN): A General Approach
Recommendations
Spatial variability of soil nutrients and GIS-based nutrient management in Yongji County, China
Integrated nutrient management is important for sustainable agricultural production and protecting environment quality and has been widely investigated around the world. In this article the spatial variability of soil nutrients was investigated and a ...
Hyperspectral vegetation indices for predicting onion (Allium cepa L.) yield spatial variability
Onion crop showed a significant yield spatial variability.Cluster analysis defined four homogeneous crop areas.Ordinary Kriging is a valuable tool to spatialize geo-referenced data.Eight vegetation indices taken are significantly related to onion ...
Study on the Spatial Variability of Farmland Soil Nutrient Based on the Kriging Interpolation
AICI '09: Proceedings of the 2009 International Conference on Artificial Intelligence and Computational Intelligence - Volume 04Studies about the spatial variability and elevation influence of soil nutrient including soil total N (TN), available P (AP) and available K (AK) could provide a theoretical basis for the scientific management of farmland and protection of ecology and ...
Comments