{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,3]],"date-time":"2026-06-03T07:27:28Z","timestamp":1780471648764,"version":"3.54.1"},"reference-count":57,"publisher":"MDPI AG","issue":"15","license":[{"start":{"date-parts":[[2021,7,24]],"date-time":"2021-07-24T00:00:00Z","timestamp":1627084800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Remote Sensing"],"abstract":"<jats:p>Understanding the biases in Deep Neural Networks (DNN) based algorithms is gaining paramount importance due to its increased applications on many real-world problems. A known problem of DNN penalizing the underrepresented population could undermine the efficacy of development projects dependent on data produced using DNN-based models. In spite of this, the problems of biases in DNN for Land Use and Land Cover Classification (LULCC) have not been a subject of many studies. In this study, we explore ways to quantify biases in DNN for land use with an example of identifying school buildings in Colombia from satellite imagery. We implement a DNN-based model by fine-tuning an existing, pre-trained model for school building identification. The model achieved overall 84% accuracy. Then, we used socioeconomic covariates to analyze possible biases in the learned representation. The retrained deep neural network was used to extract visual features (embeddings) from satellite image tiles. The embeddings were clustered into four subtypes of schools, and the accuracy of the neural network model was assessed for each cluster. The distributions of various socioeconomic covariates by clusters were analyzed to identify the links between the model accuracy and the aforementioned covariates. Our results indicate that the model accuracy is lowest (57%) where the characteristics of the landscape are predominantly related to poverty and remoteness, which confirms our original assumption on the heterogeneous performances of Artificial Intelligence (AI) algorithms and their biases. Based on our findings, we identify possible sources of bias and present suggestions on how to prepare a balanced training dataset that would result in less biased AI algorithms. The framework used in our study to better understand biases in DNN models would be useful when Machine Learning (ML) techniques are adopted in lieu of ground-based data collection for international development programs. Because such programs aim to solve issues of social inequality, MLs are only applicable when they are transparent and accountable.<\/jats:p>","DOI":"10.3390\/rs13152908","type":"journal-article","created":{"date-parts":[[2021,7,25]],"date-time":"2021-07-25T22:07:00Z","timestamp":1627250820000},"page":"2908","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":11,"title":["Bias in Deep Neural Networks in Land Use Characterization for International Development"],"prefix":"10.3390","volume":"13","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-3867-4292","authenticated-orcid":false,"given":"Do-Hyung","family":"Kim","sequence":"first","affiliation":[{"name":"Office of Innovation, UNICEF, New York, NY 10017, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Guzm\u00e1n","family":"L\u00f3pez","sequence":"additional","affiliation":[{"name":"Tryolabs, Montevideo 11300, Uruguay"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Diego","family":"Kiedanski","sequence":"additional","affiliation":[{"name":"Tryolabs, Montevideo 11300, Uruguay"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Iyke","family":"Maduako","sequence":"additional","affiliation":[{"name":"Office of Innovation, UNICEF, New York, NY 10017, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Braulio","family":"R\u00edos","sequence":"additional","affiliation":[{"name":"Tryolabs, Montevideo 11300, Uruguay"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Alan","family":"Descoins","sequence":"additional","affiliation":[{"name":"Tryolabs, Montevideo 11300, Uruguay"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Naroa","family":"Zurutuza","sequence":"additional","affiliation":[{"name":"Office of Innovation, UNICEF, New York, NY 10017, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Shilpa","family":"Arora","sequence":"additional","affiliation":[{"name":"Office of Innovation, UNICEF, New York, NY 10017, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Christopher","family":"Fabian","sequence":"additional","affiliation":[{"name":"Office of Innovation, UNICEF, New York, NY 10017, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2021,7,24]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"436","DOI":"10.1038\/nature14539","article-title":"Deep learning","volume":"521","author":"Lecun","year":"2015","journal-title":"Nature"},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"549","DOI":"10.1109\/LGRS.2017.2657778","article-title":"Training deep convolutional neural networks for land-cover classification of high-resolution imagery","volume":"14","author":"Scott","year":"2017","journal-title":"IEEE Geosci. Remote Sens. Lett."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"2217","DOI":"10.1109\/JSTARS.2019.2918242","article-title":"Eurosat: A novel dataset and deep learning benchmark for land use and land cover classification","volume":"12","author":"Helber","year":"2019","journal-title":"IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens."},{"key":"ref_4","doi-asserted-by":"crossref","unstructured":"Kussul, N., Shelestov, A., Lavreniuk, M., Butko, I., and Skakun, S. (2016, January 10\u201315). Deep Learning Approach for Large Scale Land Cover Mapping Based on Remote Sensing Data Fusion. Proceedings of the 2016 IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Beijing, China.","DOI":"10.1109\/IGARSS.2016.7729043"},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"129","DOI":"10.1016\/j.rse.2019.04.014","article-title":"Understanding urban landuse from the above and ground perspectives: A deep learning, multimodal solution","volume":"228","author":"Srivastava","year":"2019","journal-title":"Remote Sens. Environ."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"166","DOI":"10.1016\/j.isprsjprs.2019.04.015","article-title":"Deep learning in remote sensing applications: A meta-analysis and review","volume":"152","author":"Ma","year":"2019","journal-title":"ISPRS J. Photogramm. Remote Sens."},{"key":"ref_7","unstructured":"Di Gregorio, A. (2005). Land Cover Classification System: Classification Concepts and User Manual: LCCS., Food and Agriculture Organization of the United Nations."},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"21","DOI":"10.1088\/1751-8113\/44\/8\/085201","article-title":"A \u201cmissing\u201d family of classical orthogonal polynomials","volume":"44","author":"Vinet","year":"2011","journal-title":"J. Phys. A Math. Theoretical."},{"key":"ref_9","doi-asserted-by":"crossref","unstructured":"Tiecke, T.G., Liu, X., Zhang, A., Gros, A., Li, N., Yetman, G., Kilic, T., Murray, S., Blankespoor, B., and Prydz, E.B. (2017). Mapping the world population one building at a time. arXiv.","DOI":"10.1596\/33700"},{"key":"ref_10","doi-asserted-by":"crossref","first-page":"82","DOI":"10.1016\/j.isprsjprs.2020.02.014","article-title":"Deep learning-based remote and social sensing data fusion for urban region function recognition","volume":"163","author":"Cao","year":"2020","journal-title":"ISPRS J. Photogramm. Remote Sens."},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"67","DOI":"10.1080\/22797254.2019.1684208","article-title":"Earth observation tools and services to increase the effectiveness of humanitarian assistance","volume":"53","author":"Lang","year":"2020","journal-title":"Eur. J. Remote Sens."},{"key":"ref_12","unstructured":"(2021, July 21). ITU and UNICEF Have Joined Forces to Connect Every School to the Internet. Available online: https:\/\/www.itu.int\/en\/ITU-D\/Initiatives\/GIGA\/Pages\/default.aspx."},{"key":"ref_13","unstructured":"Yi, Z., Zurutuza, N., Bollinger, D., Garcia-Herranz, M., and Kim, D. (2019, January 16\u201320). Towards equitable access to information and opportunity for all: Mapping schools with high-resolution Satellite Imagery and Machine Learning. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, Long Beach, CA, USA."},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"30096","DOI":"10.1073\/pnas.1912790117","article-title":"Algorithms as discrimination detectors","volume":"117","author":"Kleinberg","year":"2020","journal-title":"Proc. Natl. Acad. Sci. USA"},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"44","DOI":"10.1145\/2447976.2447990","article-title":"Discrimination in online ad delivery","volume":"56","author":"Sweeney","year":"2013","journal-title":"Commun. ACM"},{"key":"ref_16","first-page":"671","article-title":"Big data\u2019s disparate impact","volume":"104","author":"Barocas","year":"2016","journal-title":"Calif. L. Rev."},{"key":"ref_17","first-page":"519","article-title":"Antidiscriminatory algorithms","volume":"70","author":"Bornstein","year":"2018","journal-title":"Ala. L. Rev."},{"key":"ref_18","unstructured":"Andersen, L. (2021, July 21). Artificial Intelligence in International Development: Avoiding Ethical Pitfalls. Available online: https:\/\/jpia.princeton.edu\/news\/artificial-intelligence-international-development-avoiding-ethical-pitfalls."},{"key":"ref_19","unstructured":"Mehrabi, N., Morstatter, F., Saxena, N., Lerman, K., and Galstyan, A. (2019). A survey on bias and fairness in machine learning. arXiv."},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"71","DOI":"10.1109\/MTS.2021.3056282","article-title":"From Artificial Intelligence Bias to Inequality in the Time of COVID-19","volume":"40","author":"Bullock","year":"2021","journal-title":"IEEE Technol. Soc. Mag."},{"key":"ref_21","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1145\/3386296.3386305","article-title":"Unintended machine learning biases as social barriers for persons with disabilitiess","volume":"125","author":"Hutchinson","year":"2020","journal-title":"ACM SIGACCESS Access. Comput."},{"key":"ref_22","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1038\/s41746-020-0288-5","article-title":"Sex and gender differences and biases in artificial intelligence for biomedicine and healthcare","volume":"3","author":"Cirillo","year":"2020","journal-title":"NPJ Digit. Med."},{"key":"ref_23","unstructured":"Shankar, S., Halpern, Y., Breck, E., Atwood, J., Wilson, J., and Sculley, D. (2017). No classification without representation: Assessing geodiversity issues in open data sets for the developing world. arXiv."},{"key":"ref_24","doi-asserted-by":"crossref","unstructured":"Aubakirova, M., and Bansal, M. (2016). Interpreting neural networks to improve politeness comprehension. arXiv.","DOI":"10.18653\/v1\/D16-1216"},{"key":"ref_25","doi-asserted-by":"crossref","unstructured":"Liu, N., Huang, X., Li, J., and Hu, X. (2018, January 19\u201323). On interpretation of network embedding via taxonomy induction. Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, London, UK.","DOI":"10.1145\/3219819.3220001"},{"key":"ref_26","unstructured":"Chollet, F. (2017). Deep Learning with Python, Simon and Schuster."},{"key":"ref_27","unstructured":"(2021, July 21). CS231n: Convolutional Neural Networks for Visual Recognition. Available online: http:\/\/cs231n.stanford.edu."},{"key":"ref_28","first-page":"1097","article-title":"Imagenet classification with deep convolutional neural networks","volume":"25","author":"Krizhevsky","year":"2021","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"ref_29","unstructured":"Simonyan, K., and Zisserman, A. (2015, January 7\u20139). Very deep convolutional networks for large-scale image recognition. Proceedings of the 3rd International Conference on Learning Representations, ICLR 2015\u2014Conference Track Proceedings (1\u201314), San Diego, CA, USA."},{"key":"ref_30","doi-asserted-by":"crossref","first-page":"107","DOI":"10.1016\/0141-0229(95)00188-3","article-title":"Deep Residual Learning for Image Recognition","volume":"19","author":"Yun","year":"2015","journal-title":"Enzym. Microb. Technol."},{"key":"ref_31","unstructured":"(2021, July 21). Imagenet: Vggnet, Resnet, Inception, and Xception with Keras. Available online: https:\/\/www.pyimagesearch.com\/2017\/03\/20\/imagenet-vggnet-resnet-inception-xception-keras\/."},{"key":"ref_32","unstructured":"Howard, A.G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M., and Adam, H. (2017). Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv."},{"key":"ref_33","first-page":"91","article-title":"Faster r-cnn: Towards real-time object detection with region proposal networks","volume":"28","author":"Ren","year":"2015","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"ref_34","unstructured":"Dai, J., Li, Y., He, K., and Sun, J. (2016). R-fcn: Object detection via region-based fully convolutional networks. arXiv."},{"key":"ref_35","doi-asserted-by":"crossref","unstructured":"Redmon, J., Divvala, S., Girshick, R., and Farhadi, A. (2016, January 26). You only look once: Unified, real-time object detection. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA.","DOI":"10.1109\/CVPR.2016.91"},{"key":"ref_36","doi-asserted-by":"crossref","unstructured":"Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S., Fu, C.-Y., and Berg, A.C. (2015). SSD: Single Shot MultiBox Detector. arXiv.","DOI":"10.1007\/978-3-319-46448-0_2"},{"key":"ref_37","unstructured":"O\u2019Shea, K., and Nash, R. (2015). An introduction to convolutional neural networks. arXiv."},{"key":"ref_38","unstructured":"(2021, July 23). A Comprehensive Guide to Convolutional Neural Networks\u2014The ELI5 Way. Available online: https:\/\/towardsdatascience.com\/a-comprehensive-guide-to-convolutional-neural-networks-the-eli5-way-3bd2b1164a53."},{"key":"ref_39","doi-asserted-by":"crossref","first-page":"108","DOI":"10.1080\/20964471.2019.1625151","article-title":"Global spatio-temporally harmonised datasets for producing high-resolution gridded population distribution datasets","volume":"3","author":"Lloyd","year":"2019","journal-title":"Big Earth Data"},{"key":"ref_40","unstructured":"Zhang, R. (2019, January 10). Making convolutional networks shift-invariant again. Proceedings of the International Conference on Machine Learning, Long Beach, CA, USA."},{"key":"ref_41","unstructured":"Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., and Antiga, L. (2019). Pytorch: An imperative style, high-performance deep learning library. arXiv."},{"key":"ref_42","first-page":"849","article-title":"On spectral clustering: Analysis and an algorithm","volume":"2","author":"Ng","year":"2002","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"ref_43","first-page":"659","article-title":"Gaussian Mixture Models","volume":"741","author":"Reynolds","year":"2009","journal-title":"Encycl. Biom."},{"key":"ref_44","doi-asserted-by":"crossref","first-page":"37","DOI":"10.1016\/0169-7439(87)80084-9","article-title":"Principal component analysis","volume":"2","author":"Wold","year":"1987","journal-title":"Chemom. Intell. Lab. Syst."},{"key":"ref_45","doi-asserted-by":"crossref","first-page":"227","DOI":"10.4028\/www.scientific.net\/AMR.951.227","article-title":"Automatic method for determining cluster number based on silhouette coefficient","volume":"951","author":"Zhou","year":"2014","journal-title":"Adv. Mater. Res."},{"key":"ref_46","first-page":"26853","article-title":"Akaike information criterion statistics","volume":"81","author":"Sakamoto","year":"1986","journal-title":"Dordr. Neth. D Reidel"},{"key":"ref_47","doi-asserted-by":"crossref","unstructured":"Shahapure, K.R., and Nicholas, C. (2020, January 6). Cluster Quality Analysis Using Silhouette Score. Proceedings of the 2020 IEEE 7th International Conference on Data Science and Advanced Analytics (DSAA), Sydney, Australia.","DOI":"10.1109\/DSAA49011.2020.00096"},{"key":"ref_48","doi-asserted-by":"crossref","first-page":"1921","DOI":"10.1016\/j.worlddev.2003.06.007","article-title":"Designing Policies to Reduce Rural Poverty and Environmental Degradation in a Hillside Zone of the Colombian Andes","volume":"31","author":"Agudelo","year":"2003","journal-title":"World Dev."},{"key":"ref_49","doi-asserted-by":"crossref","first-page":"185","DOI":"10.1016\/S0034-4257(01)00295-4","article-title":"Status of land cover classification accuracy assessment","volume":"80","author":"Foody","year":"2002","journal-title":"Remote Sens. Environ."},{"key":"ref_50","doi-asserted-by":"crossref","first-page":"111593","DOI":"10.1016\/j.rse.2019.111593","article-title":"Scale Sequence Joint Deep Learning (SS-JDL) for land use and land cover classification","volume":"237","author":"Zhang","year":"2020","journal-title":"Remote Sens. Environ."},{"key":"ref_51","doi-asserted-by":"crossref","unstructured":"Selvaraju, R.R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., and Batra, D. (2020, January 14). Grad-cam: Visual explanations from deep networks via gradient-based localization. Proceedings of the IEEE International Conference on Computer Vision, Seattle, WA, USA.","DOI":"10.1007\/s11263-019-01228-7"},{"key":"ref_52","unstructured":"Pesaresi, M., Ehrilch, D., Florczyk, A.J., Freire, S., Julea, A., Kemper, T., Soille, P., and Syrris, V. (2015). GHS Built-Up Grid, Derived from Landsat, Multitemporal (1975, 1990, 2000, 2014), European Commission, Joint Research Centre. JRC Data Catalogue."},{"key":"ref_53","doi-asserted-by":"crossref","first-page":"267","DOI":"10.1007\/BF02289263","article-title":"Who belongs in the family?","volume":"18","author":"Thorndike","year":"1953","journal-title":"Psychometrika"},{"key":"ref_54","doi-asserted-by":"crossref","first-page":"75","DOI":"10.1016\/j.ecolecon.2005.03.007","article-title":"Mapping regional economic activity from night-time light satellite imagery","volume":"57","author":"Doll","year":"2006","journal-title":"Ecol. Econ."},{"key":"ref_55","doi-asserted-by":"crossref","unstructured":"Engstrom, R., Pavelesku, D., Tanaka, T., and Wambile, A. (2019, January 22). Mapping Poverty and Slums Using Multiple Methodologies in Accra, Ghana. Proceedings of the 2019 Joint Urban Remote Sensing Event (JURSE), Vannes, France.","DOI":"10.1109\/JURSE.2019.8809052"},{"key":"ref_56","unstructured":"Geirhos, R., Rubisch, P., Michaelis, C., Bethge, M., Wichmann, F.A., and Brendel, W. (2018). ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness. arXiv."},{"key":"ref_57","doi-asserted-by":"crossref","first-page":"e1249","DOI":"10.1002\/widm.1249","article-title":"Ensemble Learning: A Survey","volume":"8","author":"Sagi","year":"2018","journal-title":"Wiley Interdiscip. Rev. Data Min. Knowl. Discov."}],"container-title":["Remote Sensing"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2072-4292\/13\/15\/2908\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,11]],"date-time":"2025-10-11T06:34:20Z","timestamp":1760164460000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2072-4292\/13\/15\/2908"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021,7,24]]},"references-count":57,"journal-issue":{"issue":"15","published-online":{"date-parts":[[2021,8]]}},"alternative-id":["rs13152908"],"URL":"https:\/\/doi.org\/10.3390\/rs13152908","relation":{},"ISSN":["2072-4292"],"issn-type":[{"value":"2072-4292","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,7,24]]}}}