RRP #1: Tomáš Mikolov on word2vec and AI research at Microsoft, Google, Facebook

Radim Řehůřek podcast 7 Comments

Episode Summary: Today I sat down with Tomáš Mikolov, my fellow Czech countryman whom most of you will know through his work on word2vec. But Tomáš has many more interesting things to say beside word2vec (although we cover word2vec too!): his beginnings with 8bit graphics and games, living in NY compared to California, AI research at Microsoft vs Google vs Facebook, the importance of focusing on the right thing and Tomáš' vision for the "roadmap towards machine intelligence".

Links & resources:

  1. Tomáš' current affiliation: Facebook AI Research (FAIR); FAIR profile, Google Scholar profile

  2. Mikolov, Joulin, Baroni: A Roadmap towards Machine Intelligence [PDF]

  3. A shorter version of that paper that Tomáš mentioned: Baroni &al: CommAI: Evaluating the first steps towards a useful general AI [PDF + github]

  4. original word2vec tool; tutorial for our Python port in gensim

  5. original fasttext tool; our gensim wrapper for Python

This was the first time podcasting, both for me and for Tomáš, so please take it easy on us! Still learning the ropes.

The podcast lives on SoundCloud for now. I'll upload it to iTunes once I figure out how that works.

Comments 7

  1. Pingback: Tomáš Mikolov on Word2vec and AI research at Microsoft, Google, Facebook [audio] | ExtendTree

  2. Piero Savastano

    Great interview!
    I hope one day Tomas will be able to make machines talk and learn in a natural way, and that Radim will help bring that power into the industry.
    Both of you guys are inspiring and worth following, good luck 🙂

  3. Kenichi SUZUKI

    Great piece of work and thank you for the episode.
    I also would like to thank you for your contribution in gensim.
    Surprised to know you are now in Geoje.
    (I’m curious how you end up in Geoje.)
    It’s funny that, I feel, the more I know about the personality of Mikolov-san from the interview, the more I like word2vec.
    Please keep up the good work on the podcasts!!

    1. Post
  4. Pingback: Learning languages, machine translation, and neural networks

  5. Mira Hedl

    Thanks for the interview.
    I remember our discussion with Tom on high school around year 2000 when he already had passion for general-purpose AI, it was before the boom and excitement and attention the topic gets these days.
    I’m glad that he is on his way to pursue that childhood vision.

    Interview was pleasant to listen to, two leading edge engineers and researchers friendly chatting. It’d be maybe good to have some structure prepared next time, but that unstructured format wasn’t a problem in this case IMO.

    I wish good luck to both of you (from Prague).
    Keep up that excellent work you are both doing!

Leave a Reply

Your email address will not be published. Required fields are marked *