Presentations

Here you can find CNTK and Deep Learning related presentations:

• 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

Recent

• 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.

• 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).

Upcoming

• 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

Past

• 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](https://cntk.ai/kdd/CNTK Hands-On KDD2016, Frank Seide and Amit Agarwal.zip); 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.