Presentations

Here you can find CNTK and Deep Learning related presentations:

Recent

  • GTC 2017 - May 8, 2017: Sayan Pathak presented Scalable deep learning with Microsoft Cognitive Toolkit. Presentation links - Overview, Tutorial.

  • Build 2017 - May 9-11, 2017: Sayan Pathak presented Deep learning with Cognitive Toolkit. Presentation, Video

  • CVPR 2017 - July 26, 2017, Tutorial by Emad Barsoum, Sayan Pathak and Cha Zhang, Scalable Deep Learning with Microsoft Cognitive Toolkit. Presentation

Past

  • WWW 2017 - April 3-7, 2017: Half-day tutorial titled "Scalable deep document (sequence) reasoning with Cognitive Toolkit". Details of the tutorial can be found here

  • AI NEXT 2017 conference: On March 17, 2017, Frank Seide presented CNTK incl. the new functional "Layers" API. Video and Slides.

  • NVidia Inception Webinar on Feb 9, 2017 where Sayan Pathak presented an overview of CNTK V2 with latest performance benchmarks as well as a introduction to deep learning with Cognitive Toolkit (formerly CNTK).

  • Keynote at CGO-HPCA-PPoPP-CC joint conference on Feb 8, 2017 where Frank Seide presented CNTK with focus on the computer science behind it. Slides are here. Video will be linked from here once available.

  • Sayan Pathak Cognitive Toolkit (CNTK) Deep Dive and Hands-on Tutorial - Nov 2016 This YouTube video goes from a high level view, to an intermediate level of detail and then to hands-on tutorials. The first half is a talk and slideware. The later half is hands-on tutorials

  • KDD 2016: Frank Seide and Amit Agarwal presented a tutorial titled, "CNTK: Microsoft's Open-Source Deep-Learning Toolkit" on August 16, 2016. Tutorial description here, Prerequisites page here; slides here; video here, note: 3 parts. This 3-hour hands-on tutorial covers the following topics:

    • what: ...is CNTK? Including an introduction into the core concept of CNTK, the "computational network."
    • how: ...does a typical use of CNTK look like? Configuration of key components, workflow.
    • deep dive: ...into unique CNTK technologies. Automatic unrolling of time & efficient minibatching of variable-length sequences; data-parallel training with 1-bit SGD and Block Momentum.
    • hands-on tutorials:
      • hello world: getting set up, running a first logistic regression task
      • language understanding: slot and intent tagging on the ATIS language-understanding corpus. Available online
      • image recognition: with convolutional nets, batch normalization, and residual networks using the CIFAR-10 corpus Available online
  • Microsoft Research Latin American Faculty Summit 2016: Frank Seide presented, "CNTK: Microsoft's Open-Source Deep-Learning Toolkit," on . May 19, 2016. Video and Slides. The 1-hour talk covers the following topics:

    • what: ...is CNTK? Including an introduction into the core concept of CNTK, the "computational network."
    • how: ...does a typical use of CNTK look like? Configuration of key components, workflow.
    • deep dive: ...into unique CNTK technologies. Automatic unrolling of time & efficient minibatching of variable-length sequences; data-parallel training with 1-bit SGD and Block Momentum.
    • examples: source code walk-through. For ResNet image classification and sequence-to-sequence modeling with attention.
  • GPU Technology Conference 2016: Alexey Kamenev presented, "Deep Learning in Microsoft with CNTK" on April 2016. Video and Slides.