Federated Deep Learning in SAAS Applications

schedule Aug 8th 01:45 - 02:30 PM place Neptune people 54 Interested

ML in Saas Applications becomes exceedingly difficult due to lack to access to customer data. The Customer Data is locked down with no outside access and it presents a huge problem to do ML on this data in a traditional way. The focus of this presentation is to provide alternate solutions to do ML in a distributed fashion.

We will focus on Split Neural Networks - a relatively new distributed ML Technique to solve Data Access issues with a SAAS Application.

We will walk through the motivations behind Split Neural Network approach to ML .

We will go through some concrete examples that are already using this technique.

 
 

Outline/Structure of the Demonstration

Machine Learning and Privacy

(Anonymous, Obfuscated, Encrypted) DATA !

Privacy and ML – Techniques

Machine Learning at Healthcare and SAAS applications

Split Neural Network – Introduction

SAAS Use Case – Anomaly Detection

Split NN to rescue

Learning Outcome

Learn the latest world of distributed deep learning techniques.

Target Audience

All Machine Learning Practitioners

Prerequisites for Attendees

Neural Network Fundamentals

schedule Submitted 4 months ago

Public Feedback

comment Suggest improvements to the Speaker
  • Dr. Vikas Agrawal
    By Dr. Vikas Agrawal  ~  4 months ago
    reply Reply

    Dear Arun: I would love to see some specific details of a case study we plan to show during the talk. That will make it very attractive for potential attendees. Warm Regards, Vikas

    • Arun Krishnaswamy
      By Arun Krishnaswamy  ~  3 months ago
      reply Reply

      Hi Vikas,

      Will send a detailed doc after removing some of the legal content. 

      Thanks

      Arun

      • Dr. Vikas Agrawal
        By Dr. Vikas Agrawal  ~  4 months ago
        reply Reply

        Dear Arun: The material we show at Open Data Science Conference needs to be freely shareable as the presentation and videos will be shared on the web. One possibility is to sanitize your materials from employer specificity and present the use case, its demo and problem solution in a way that is specific enough to get the mathematics and technical content across without infringing on IP. In other words, please share it in a way that is acceptable for sharing as a representative of your employer on a public website. Warm Regards, Vikas

        • Arun Krishnaswamy
          By Arun Krishnaswamy  ~  3 months ago
          reply Reply

          Hi Vikas,

          That's precisely the process I'm working on.  Will share the newly updated deck at the earliest. I already shared an initial doc with you over LinkedIn ( I hope u got that ).

          Thanks

          Arun

          • Dr. Vikas Agrawal
            By Dr. Vikas Agrawal  ~  3 weeks ago
            reply Reply

            Dear Arun: The initial presentation you shared was Praneeth's talk on Split NNs. As discussed, we will add him as a co-author and co-presenter.

            Warm Regards

            Vikas

          • Dr. Vikas Agrawal
            By Dr. Vikas Agrawal  ~  3 months ago
            reply Reply

            Dear Arun: Are you in a position to share the details of the work after permission from your collaborators, please? Warm Regards, Vikas

            • Naresh Jain
              By Naresh Jain  ~  3 months ago
              reply Reply

              Hi Arun,

              Can you please update us on the status? The topic is interesting, however without concrete details, the program committee won't be in a position to select the proposal.

              We are doing a final round of selection. If you can share the details in the next couple of days, it would really be helpful.