[2407.15339] Deep Learning for Economists

View a PDF of the paper titled Deep Studying for Economists, by Melissa Dell

View PDF
HTML (experimental)

Summary:Deep studying supplies highly effective strategies to impute structured info from large-scale, unstructured textual content and picture datasets. For instance, economists would possibly want to detect the presence of financial exercise in satellite tv for pc photographs, or to measure the matters or entities talked about in social media, the congressional document, or agency filings. This evaluate introduces deep neural networks, masking strategies similar to classifiers, regression fashions, generative AI, and embedding fashions. Functions embrace classification, doc digitization, document linkage, and strategies for information exploration in huge scale textual content and picture corpora. When appropriate strategies are used, deep studying fashions may be low cost to tune and might scale affordably to issues involving thousands and thousands or billions of knowledge factors.. The evaluate is accompanied by a companion web site, EconDL, with user-friendly demo notebooks, software program assets, and a data base that gives technical particulars and extra functions.