Microsoft Research provides a continuously refreshed collection of free datasets, tools and resources designed to advance the state of the art of academic research in many areas of computer science, such as natural language processing and computer vision. In addition, you can browse datasets and apply for cloud-based compute cycles available under the Azure for Research program.
Submissions were received from around the world, and the judging panel selected the projects that best show how Microsoft AI services are used to solve a difficult problem and provide insightful solutions. Congratulations to the winners, and thanks to all who submitted a project for the Cloud AI Research Challenge!
>>Click here to view the winners.
Microsoft Research is proud to support the US-wide National Science Foundation’s (NSF) Big Data Regional Innovation Hubs (BD Hubs) program by awarding $3M in Microsoft Azure cloud computing credits.
NSF supports four regional hubs for data science innovation, called the BD Hubs, throughout the United States. The consortia are coordinated by top data scientists at multitude of top universities around the country.
Program Director
Members of the Data science team were recently at these events:
(29.8 GB, August 2015)
This heterogeneous graph is composed of six entity types modeling scholarly activities (count): Field of Study (50K), Author (20M), Institutions (770K), Venue, Paper (83M), and Event (26K).
Use cases: Improving statistical rankings, natural language search, question-answering
(279.02 MB, January 2016)
A full census of 38 million tweet IDs collected for social media analysis from July 2012 to November 2012.
Use cases: Understanding of topic-specific interest and opinion amongst the general population and specific demographic subgroups with regards to political issues
(March 31, 2016)
A series of actual versus prediction datasets for windflow at 3000, 6000, 9000, 12000, 18000, 24000, 30000, 34000, 39000, 45000, and 53000 elevations. The site also shows a visualization of the both the actual NOAA data versus the predictive values.
Use cases: Future prediction of windflow patterns
(139.42MB, October 2015)
This dataset contains knowledge base relation triples and textual mentions of Freebase entity pairs.
Use cases: can be used to obtain a higher accuracy on knowledge base completion by disambiguating between closely synonymous textual relations that share common sub-structure, and are composed of similar words and dependency arcs.
(2048MB, August 2015)
This dataset contains video-sequences of several individuals performing hand gestures, as captured by a depth camera. The ground truth locations of the fingertips are included as an annotation for each frame of the video.
Use Cases: can be used to create or verify hand tracking systems
Datasets are ordered by recency (more recent at the top) and by their primary research area. If you don’t find what you are looking for in one area, be sure to browse the others.
These datasets have to do with methods of interactions between humans and computers such as gesture, speech, and touch.
These datasets are video, continuous images, or audio files.
These datasets are semantic maps or image-retrieval data.
These datasets contain GPS traces for a variety of applications; validation, activity analysis, and location obfuscation.
These datasets cut across various NLP applications from semantic analysis, entity extraction, captioning (of video), question answering and many others.
These datasets are designed to be used for image and video recognition captured by still cameras, depth cameras, robots, and standard video. These clips and images usually include tags or captions to assist with training recognition systems.
