Commercial License RRS feed

  • Pergunta

  • I want to apply for a commercial license to post game plays in youtube what all is required and terms related to it .
    domingo, 15 de janeiro de 2017 12:05

Todas as Respostas

  • If you want a commercial license for the Infer.NET library, send email to infersup@microsoft.com.
    domingo, 15 de janeiro de 2017 16:03
  • Its hard to seriously consider this interesting software due to restrictive licensing.

    I am heavy user of MS CNTK which is open source and free for commercial use.

    Maybe under the new MS spirit of embracing open source, as embodied by CNTK, this software will also be made more freely available.

    For now, maybe the better option is to use Stan (http://mc-stan.org/)  

    quarta-feira, 13 de junho de 2018 12:37
  • Stan is a great tool with many state-of-the-art algorithms for Bayesian statistics. Infer.NET features great and mature inference algorithms as well. With Infer.NET, you can easily build models that can be naturally expressed as factor graphs. You can't do that in Stan. The two frameworks are not direct competitors in my opinion. They have slightly different audiences.

    In Stan you can't have discrete random variables. Authors argue that these can be taken care using marginalization but for new users and beginners in Bayesian statistics this is not that easy.

    To my knowledge, Infer.NET is the only mature probabilistic programming framework that supports true online learning. That is, you can use the posterior as prior in the next iterations. There are some limitations and technical details but I managed to get that working in my models. This is very useful when data cannot fit in memory. Stan and PyMC3 don't support online learning.

    • Editado usptact sábado, 23 de junho de 2018 08:32
    sábado, 23 de junho de 2018 08:30
  • We're working on this. I can't yet promise anything, but would expect to have an update in the next 3-4 months.


    segunda-feira, 25 de junho de 2018 14:59
  • Infer.NET is now released as open source under the MIT license.
    sexta-feira, 5 de outubro de 2018 14:37