Results 1 to 2 of 2

Thread: Intel researchers compress AI models without compromissing accuracy

  1. #1
    Member
    Join Date
    Dec 2008
    Location
    Poland, Warsaw
    Posts
    909
    Thanks
    531
    Thanked 359 Times in 267 Posts

    Intel researchers compress AI models without compromissing accuracy

    https://venturebeat.com/2019/06/11/i...sing-accuracy/

    Hmmm. I wonder if it's really something new...

  2. #2
    Administrator Shelwien's Avatar
    Join Date
    May 2008
    Location
    Kharkov, Ukraine
    Posts
    3,134
    Thanks
    179
    Thanked 921 Times in 469 Posts
    That post links to https://arxiv.org/abs/1902.0596

    > beginning with a compact model and modifying the structure based on data during training.
    > They claim it’s more scalable and computationally efficient than starting with a large model followed by compression,
    > because training operates directly on the compact model.

    I think the idea is to directly train a compressed (like RLE) representation of a large sparse model.

Similar Threads

  1. Justification for data compression to be very foundation of AI
    By LawCounsels in forum Data Compression
    Replies: 81
    Last Post: 9th April 2019, 09:21
  2. Researchers Break Thousands of RSA Keys
    By Sportman in forum The Off-Topic Lounge
    Replies: 0
    Last Post: 16th January 2019, 02:21
  3. Replies: 45
    Last Post: 25th November 2016, 04:30
  4. About PAQ8 Models Training
    By masigabriel in forum Data Compression
    Replies: 4
    Last Post: 10th October 2016, 20:49
  5. Researchers warn of malware hidden in .zip files
    By Surfer in forum The Off-Topic Lounge
    Replies: 4
    Last Post: 20th April 2010, 09:19

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •