Research
Discover an index of datasets, SDKs, APIs and open-source tools developed by Microsoft researchers and shared with the global academic community below. These experimental technologies—available through Azure AI Foundry Labs (opens in new tab)—offer a glimpse into the future of AI innovation.
The Orthant-Wise Limited-memory Quasi-Newton algorithm (OWL-QN) is a numerical optimization procedure for finding the optimum of an objective of the form {smooth function} plus {L1-norm of the parameters}. It has been used for training log-linear…
HTML and JavaScript Web code that displays a user interface to enable users to search, assimilate, and compare maps rapidly from an extensive corpus. The download includes a reference to a sample corpus of approximately…
Results from experiments conducted by Microsoft Research’s Machine Translation Incubation Team to investigate the impact of using good English (controlled language) on post-editing productivity—as well as on the overall quality of our statistical machine-translation system.
A set of manually annotated Chinese word-segmentation data and specifications for training and testing a Chinese word-segmentation system for research purposes. The data was extracted from the People’s Daily, which we have licensed for commercial…
This tool enables training and testing of maximum-entropy models using a general-feature file format. The tool also supports RProp and GIS as training algorithms.
MapCruncher lets users quickly convert existing maps into an online format that’s as fast and easy to use as Virtual Earth. PDF and raster maps can be converted in minutes just by clicking on corresponding…