Anthill Inside 2017

On theory and concepts in Machine Learning, Deep Learning and Artificial Intelligence. Formerly Deep Learning Conf.

From RNN to Attention

Submitted by Sarath R Nair (@s4sarath) on Monday, 12 June 2017

videocam_off

Technical level

Intermediate

Section

Full talk

Status

Submitted

Vote on this proposal

Login to vote

Total votes:  +1

Abstract

The motivation and ideas behind RNN to LSTM to Attention Mechanism and wind up with latest trends , along with mathematical ideas of Backpropogation .

Outline

The idea of backpropogation remains a mystery to many people and especially to those who wants to gain some intuition about , deriving it by themselves . This talk , focuss on the basic ideas of BackProp and how one can easily extend this to RNN and LSTM . This will be followed by Attention Mechanism in LSTM and current state of the art techniques in Tensorflow .

Requirements

Basic Knowledge of Deep learning , and Natural Language Processing

Speaker bio

I am Sarath R Nair , working as a Data Scientist from past 2 years. I started exploring Deep Learning by 2014, and now I am mainly focussing on the maths and core concepts behind DL, along with practical experience in DL for NLP , basic Data Modelling and hands on experience in BigData Stack .

Comments

  • 1
    Sandhya Ramesh (@sandhyaramesh) Reviewer a year ago

    Hi Sarath, in order to evaluate your proposal, please ensure you submit a slide deck and a two minute self recorded video walking us through your talk.

    • 1
      Sarath R Nair (@s4sarath) Proposer a year ago

      Hi , I am in the process of making it. I need some more time ( appx a week ) . For the time being, if you want to look at a draft, here is the link . https://www.slideshare.net/secret/HZqELeefsFtmuZ . The final slide will be entirely different from this.

      • 1
        Sandhya Ramesh (@sandhyaramesh) Reviewer a year ago

        Hi Sarath, please submit your slides and video before Friday the 22nd. This is our hard deadline for this proposal.

Login with Twitter or Google to leave a comment