Deep Learning Architectures- A Deep Dive

Description:

This workshop will provide you a detailed idea of important developments in the field of computer vision and convolutional neural networks. We’ll look at some of the most important architectures over the last 5 years and discuss why they’re so important. I will take you through the basics of how a Convolutional Network works and then we'll step by step delve deeper into the various architectures built around it. You'll learn how to implement the architectures in Keras and also how to use them for your dataset. I'll also take you through the basics of Transfer learning and help you understand how to code for the same in python (Keras).

Below is a rough outline of the topics I plan to cover during the workshop

Outline

1. Introduction

• What is a Convolutional Neural Network
• What is the difference between a CNN and a Neural Network

2. CNN Architectures

• Understanding various layers of a CNN architecture

• Understanding and Learning how to calculate the input and output volume size and parameters of various layers (DIY on VGG Net)

• Implementation of the architectures (AlexNet and VGG Net)

• DIY (Implement ZFNet yourself)

• Introduction and Implementation of Transfer Learning

3. Evolution of Various Architectures and the differences between them(GoogleNet, ZF Net, ResNet)

4.Q&A

Time Break-Up

1. Introduction -> 20 minutes
2. CNN Architectures -> 110 minutes (15 + 25 + 30 + 20 + 20 )
3. Architectures-> 20 minutes
4. Q&A 10 minutes

Prerequisites:

• Python3 Installed
  • Jupyter Installed
  • Basic knowledge of Neural Networks
  • Interest towards Deep Learning (CNN)

Content URLs:

In the process of making content. The contents for hands on the session can be found in this repository.

Speaker Info:

I am currently a Software Engineer at SAP Labs India working on building intelligent solutions for Quality Assurance. I am a Machine Learning Enthusiast and have been working on Machine and Deep Learning for almost a year. I have been a constant admirer of ML/DL advancements and frameworks.

Speaker Links:

Github

Linkedin

Medium