人口資料於各學科與領域皆有使用上需求，其中地理學注重於討論人口資料於空間上分佈位置。近年來隨著政府資料公開，Open Data可取得人口資料最精細尺度為最小統計區，但台灣政府受限於法治規定，無法開放戶籍門牌尺度人口資料。更精細的人口資料可以減少人口推估誤差，一直以來都有此需求。
近年來由於電腦硬體技術提升，使深度學習理論再次受到重視與使用。近期人口推估研究也開始使用深度學習理論進行人口推估。
本研究使用分區密度法，多層感知器與卷積神經網路，分別建立三種人口推估模型。並使用容積率、建蔽率、樓地板面積樓層高度、建物型態、國土利用調查成果圖、都市計畫土地使用分區圖等資料做為模型訓練因子，最終產製出5公尺人口網格資料，並與戶籍人口資料進行驗證比對。
研究結果顯示卷積神經網路人口推估模型推估結果最為優秀，模型訓練表現優於多層感知器人口推估模型，卷積神經網路人口推估模型Adjusted R2可達0.72585。採用深度學習方法人口推估模型與採用傳統方法人口推估模型相比，更不容易出現極端人口高估與低估現象。

Population data are used in various disciplines and fields. Geography focuses on discussing the spatial distribution of population data. In recent years, with the disclosure of government data, Open Data can obtain population data at the most precise scale, which is the smallest statistical area. However, the Taiwan government is restricted by legal regulations and cannot open up population data at the household registration number level. There is an ongoing need for more granular demographic data to reduce errors in population estimates.
In recent years, due to the improvement of computer hardware technology, deep learning theory has once again been valued and used. Recent population estimation research has also begun to use deep learning theory for population estimation.
This study uses the partition density method, multi-layer perceptron and convolutional neural network to establish three population estimation models. And use data such as floor area ratio, built-up coverage ratio, floor area and floor height, building type, land use survey results map, urban planning land use zoning map and other data as model training factors, and finally produce a 5-square-meter population grid data , and verify and compare with household registration population information.
Research results show that the convolutional neural network population estimation model has the best estimation results. The model training performance is better than the multi-layer perceptron population estimation model. The adjusted R2 of the convolutional neural network population estimation model can reach 0.72585. Compared with population estimation models using traditional methods, population estimation models using deep learning methods are less prone to extreme population overestimation and underestimation.

李宏毅（2016/03/21）：〈[DSC 2016] 系列活動：李宏毅 / 一天搞懂深度學習〉。slideshare。https://www.slideshare.net/tw_dsconf/ss-62245351（2023/10/20瀏覽）。
林美君. (2011). 多層多類分區密度之空間人口重分布模式. 國立臺灣大學. Available from Airiti AiritiLibrary database. (2011年)。
林庭嘉. (2021). 不同尺度空間插值法進行人口分佈推估的比較研究. (碩士), 國立臺灣師範大學, 台北市. Retrieved from https://hdl.handle.net/11296/xqwqy4。
株式会社アイデミー, 山口達輝, 松田洋之（2021）：《圖解AI｜機器學習和深度學習的技術與原理 (電子書)》（衛宮紘譯）。碁峰。（原著出版年2019）。
鄧志松（2015/02/05）：〈5.5 地圖計算：空間內插〉。Excel2Earth與空間分析。http://excel2earth.blogspot.com/（2022/04/06瀏覽）。
謝心怡. (2007). 多層多類別之人口地理分布模式. 國立臺灣大學. Available from Airiti AiritiLibrary database. (2007年)。
周敬棋、林先和、陳正誠、陳潤秋、詹大千.（2022）.〈比較靜態與動態人口資料應用於新冠肺炎熱區之預測能力〉。《台灣公共衛生雜誌》，41。611-626。
Cheng, L., Wang, L., Feng, R., & Yan, J. (2021). Remote Sensing and Social Sensing Data Fusion for Fine-Resolution Population Mapping With a Multimodel Neural Network. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 14, 5973-5987. doi: 10.1109/JSTARS.2021.3086139.
Comber, A., & Zeng, W. (2019). Spatial interpolation using areal features: A review of methods and opportunities using new forms of data with coded illustrations. Geography Compass, 13(10). doi: 10.1111/gec3.12465.
Da Silva, I. N., Hernane Spatti, D., Andrade Flauzino, R., Liboni, L. H. B., & dos Reis Alves, S. F. (2017). Artificial Neural Network Architectures and Training Processes Artificial Neural Networks (pp. 21-28).
Douglass, R. W., Meyer, D. A., Ram, M., Rideout, D., & Song, D. (2015). High resolution population estimates from telecommunications data. EPJ Data Science, 4(1). doi: 10.1140/epjds/s13688-015-0040-6.
Doupe, P., Bruzelius, E., Faghmous, J., & Ruchman, S. G. (2016). Equitable development through deep learning: The case of sub-national population density estimation. Paper presented at the Proceedings of the 7th Annual Symposium on Computing for Development, Nairobi, Kenya. https://doi.org/10.1145/3001913.3001921.
Georgati, M., Monteiro, J., Martins, B., & Keßler, C. (2022). Spatial Disaggregation of Population Subgroups Leveraging Self-Trained Multi-Output Gradient Boosting Regression Trees. AGILE: GIScience Series, 3, 1-14. doi: 10.5194/agile-giss-3-5-2022.
John E. Ball, Derek T. Anderson, & Chee Seng Chan. (2017). A Comprehensive Survey of Deep Learning in Remote Sensing: Theories, Tools and Challenges for the Community. Journal of Applied Remote Sensing, 11(4).
Lam, N. S.-N. (2013). Spatial Interpolation Methods: A Review. The American Cartographer, 10(2), 129-150. doi: 10.1559/152304083783914958.
Langford, M. (2013). An Evaluation of Small Area Population Estimation Techniques Using Open Access Ancillary Data. Geographical Analysis, 45(3), 324-344. doi: 10.1111/gean.12012.
Li, T., Pullar, D., Corcoran, J., & Stimson, R. (2007). A comparison of spatial disaggregation techniques as applied to population estimation for South East Queensland (SEQ), Australia. Applied GIS, 3(9), 16. doi: 10.4225/03/57E9AECEBA789.
Lin, J., & Cromley, R. G. (2015). Evaluating geo-located Twitter data as a control layer for areal interpolation of population. Applied Geography, 58, 41-47. doi: 10.1016/j.apgeog.2015.01.006.
Liu, X. H., Kyriakidis, P. C., & Goodchild, M. F. (2008). Population‐density estimation using regression and area‐to‐point residual kriging. International Journal of Geographical Information Science, 22(4), 431-447. doi: 10.1080/13658810701492225.
Monteiro, J., Martins, B., & Pires, J. M. (2017). A hybrid approach for the spatial disaggregation of socio-economic indicators. International Journal of Data Science and Analytics, 5(2-3), 189-211. doi: 10.1007/s41060-017-0080-z.
Monteiro, J., Martins, B., Costa, M., & Pires, J. M. (2021). Geospatial Data Disaggregation through Self-Trained Encoder–Decoder Convolutional Models. ISPRS International Journal of Geo-Information, 10(9). doi: 10.3390/ijgi10090619.
Monteiro, J., Martins, B., Costa, M., & Pires, J. M. (2022). A co-training approach for spatial data disaggregation. Paper presented at the Proceedings of the 30th International Conference on Advances in Geographic Information Systems, Seattle, Washington. https://doi.org/10.1145/3557915.3561475.
Monteiro, J., Martins, B., Murrieta-Flores, P., & Pires, J. M. (2019). Spatial Disaggregation of Historical Census Data Leveraging Multiple Sources of Ancillary Information. ISPRS International Journal of Geo-Information, 8(8), 327.
Nagle, N. N., Buttenfield, B. P., Leyk, S., & Speilman, S. (2014). Dasymetric Modeling and Uncertainty. Ann Assoc Am Geogr, 104(1), 80-95. doi: 10.1080/00045608.2013.843439.
O'Shea, K., & Nash, R. (2015). An Introduction to Convolutional Neural Networks. CoRR, abs/1511.08458.
Qiu, Y., Zhao, X., Fan, D., & Li, S. (2019). Geospatial Disaggregation of Population Data in Supporting SDG Assessments: A Case Study from Deqing County, China. ISPRS International Journal of Geo-Information, 8(8). doi: 10.3390/ijgi8080356.
Robinson, C., Hohman, F., & Dilkina, B. (2017). A Deep Learning Approach for Population Estimation from Satellite Imagery. Paper presented at the Proceedings of the 1st ACM SIGSPATIAL Workshop on Geospatial Humanities.
Stevens, F. R., Gaughan, A. E., Linard, C., & Tatem, A. J. (2015). Disaggregating census data for population mapping using random forests with remotely-sensed and ancillary data. PLoS One, 10(2), e0107042. doi: 10.1371/journal.pone.0107042.
Tiebei, L., David, P., Jonathan, C., & Robert, S. (2007). A comparison of spatial disaggregation techniques as applied to population estimation for South East Queensland (SEQ), Australia. Applied GIS, 3:1–16.
Tobler, W. R. (1979). Smooth pycnophylactic interpolation for geographical regions. J Am Stat Assoc, 74(367), 519-530. doi: 10.1080/01621459.1979.10481647.
Wu, T., Luo, J., Dong, W., Gao, L., Hu, X., Wu, Z., . . . Liu, J. (2020). Disaggregating County-Level Census Data for Population Mapping Using Residential Geo-Objects With Multisource Geo-Spatial Data. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 13, 1189-1205. doi: 10.1109/jstars.2020.2974896.