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.
This code supports the paper “Deep Probabilistic Logic: A Unifying Framework for Indirect Supervision”, presented at EMNLP 2018.
Machine learning models for edge devices need to have a small footprint in terms of storage, prediction latency, and energy. One example of a ubiquitous real-world application where such models are desirable is resource-scarce devices…
Ambrosia is a programming language independent approach for authoring and deploying highly robust distributed applications. Ambrosia dramatically lowers development and deployment costs and time to market by automatically providing recovery and high availability.
This download contains the processed datasets used in the AAAI’19 paper ABC: Efficient Selection of Machine Learning Configuration on Large Dataset. A machine learning configuration refers to a combination of preprocessor, learner, and hyperparameters. Given…
The reactive machine is a programming model for microservices that provides virtual resiliency and elasticity: programmers can focus on the application logic, while the runtime transparently handles machine failures, connection failures, and distribution over a…
Microsoft Simple Encrypted Arithmetic Library (Microsoft SEAL) is an easy-to-use homomorphic encryption library developed by researchers in the Cryptography Research group at Microsoft Research. SEAL is written in modern standard C++ and has no external…
Because Microsoft Genomics is on Azure, you have the performance and scalability of a world-class supercomputing center, on demand in the cloud. Take advantage of a backend network with MPI latency under three microseconds and…
ReDial (Recommendation Dialogues) is an annotated dataset of dialogues, where users recommend movies to each other. The dataset was collected by a team of researchers working at Polytechnique Montréal (opens in new tab), MILA –…
Synthetic controls is the most common empirical strategy used to accommodate unobserved endogenous factor models. However, little consideration is given to the dimension of the relevant factors, whether the associated loadings are truly time-invariant and…
Implementation of TACL 2017 paper: “Cross-Sentence N-ary Relation Extraction with Graph LSTMs.” Nanyun Peng, Hoifung Poon, Chris Quirk, Kristina Toutanova and Wen-tau Yih.