Chenlin Meng

Our society is facing formidable sustainability challenges with more than 700 million people still living in extreme poverty. Progress towards many United Nations Sustainable Development Goals (SDGs) is hampered by a persistent lack of key socio-economic data, particularly in developing countries. For instance, data on poverty -- the first of seventeen SDGs -- is both spatially sparse and infrequently collected due to the high cost of on-the-ground surveys. Machine learning (ML) techniques hold great promise to close the data gap in many sustainability applications. Motivated by real-world problems in sustainability applications, I aim to incorporate domain knowledge from earth system science and economics into ML algorithms to improve the robustness and effectiveness of ML algorithms in sustainability applications. With such algorithms, globally available and passively collected non-traditional data sources can be used to provide insights into progress toward SDGs.