Academic Torrents: A Distributed System For Sharing Enormous Datasets.

Aug. 18, 2018, 8:53 p.m. By: Kirti Bakshi

academic torrents

As competitions are getting more popular day by day, the transfer of increasingly larger data sets has become not only infeasible but costly as well. ImageNet has since become popular beyond the competition, and now has many papers and models revolving around this data set but in order to share such an important resource to the community of machine learning, the sharers of ImageNet must shoulder the burden of a large bandwidth.

Academic Torrents, for disseminating competition data reduces this burden, and for end users also increase download speeds. With a peer-to-peer swarm, by the augmentation of an existing HTTP server, requests get re-routed to get data from downloaders. While existing systems with more users slow down, the benefits of Academic Torrents, with noticeable effects just grow even when only one other person is downloading.

What is Academic Torrents?

Academic Torrents, maintained by the pending-nonprofit Institute for Reproducible Research is a community-maintained distributed repository for datasets and scientific knowledge.

Academic Torrents is a distributed system that is meant for the sharing of enormous datasets - for researchers, by researchers. The result is a fault-tolerant repository for data that is scalable, secure and gives the benefit of fast download at blazing speeds.

About People:

  • Joseph Paul Cohen - Founder/Board Member

  • Henry Z Lo - Founder/Board Member

  • Jonathan Nogueira - Founder/Board Member

Academic Torrents team


This service has been designed in order to facilitate storage of all the data that is made use of in research, that includes datasets as well as publications. Given below are a few among the many advantages of using bittorrent technology to disseminate this work.

  • The provision of Distributed storage and content delivery by anyone. Files can be securely downloaded from other users of the system. The sharing of a file can be for a day or a year.

  • Mirroring the content can be done from a desktop computer anywhere. Everyone that surrounds this computer will automatically and securely have local access to the data.

  • Bundles of files, not just papers, or any size can be disseminated in this way as long as at least one person can become a seed for that data.

Torrent technology, with just a torrent client, allows a group of editors to "seed" their own peer-reviewed published articles. Each editor can have a part or all of the papers stored on their desktops and without a dedicated server have a torrent tracker to coordinate the delivery of papers.
One aim of this site is the creation of the infrastructure in order to allow open access journals to operate at low cost. By facilitating file transfers, the journal can focus on its core mission of providing research of world-class standards.

Large dataset delivery can be supported by researchers in the field that have the dataset present on their machine. A popular large dataset doesn't need to be centrally housed as Researchers can have part of the dataset they are working on and they can help host it together. And Libraries, without becoming the only source of the data can host this data to host papers from their own campus.

Academic Torrents is currently making 27.31TB of research data available and the future efforts will help scale up the platform!

How To Download Academic Torrents:

More Information: Click Here