Automation loop for model training using Python

VikramSridhar


Description:

I am proposing an idea to bring model training pipeline inside an automation loop. The example I am going to discuss is an NLP based classification use case.For this use case ML engineers have to try different hyper parameters and find an optimal one before shipping a model. Also ML engineers have to understand the model failures due to data uncertainty and manipulate the training data if required. These analysis will be performed by the solution I am proposing and take care of model training inside the automation loop with new configuration for each training loop

You will learn:

  • Machine Learning Pipeline - 5 Minutes
  • Data uncertainty and Model uncertainty - 5 Minutes
  • How to automate a model training inside a loop- 15 minutes
  • Q&A - 5 minutes

Prerequisites:

  • Basic understanding of Machine learning model training and optimisation
  • Deep learning understanding
  • Machine Learning pipeline

Video URL:

https://youtu.be/gZwWzdX8zdU

Content URLs:

Presentation is present at :https://gitlab.com/tcin/pyconindia2021/

Speaker Info:

I am Vikram Sridhar. I am working in Toyota Connected India as a Machine Learning Engineer. I primarily work on deep learning NLP/CV use cases.I have around 7+ years of experience in IT industry. I have great interest in working for AI/IOT based products and build a shippable and scalable solution to the end users.

Section: Data Science, Machine Learning and AI
Type: Talks
Target Audience: Intermediate
Last Updated: