The Advent of Deep Neural Networks. Neural Network implementation without ML libraries and extending them with Tensorflow.

Aniket Chowdhury (~aniket43)




The focus is more on teaching core concepts to programmers rather than using libraries. More than one neural network will be implemented.

Neural Networks

An Easy way to learn Machine Learning

An interactive way to learn ML. With ML being a leading platform in the market, the workshop introduces to one of the most important fields of Machine Learning that is Deep Neural Networks. Only basic introduction to Mathematics required.

Why Python?

  • Python for Machine Learning

Machine Learning

  • What is Machine Learning?
  • Why learn Machine Learning?
  • Types of Machine Learning
  • Regression and Classification
  • Supervised and Unsupervised

Neural Networks

  • Deep Neural Networks
  • Feed forward Neural Networks
  • Convolutional Neural Networks CNN
  • Recurrent Neural Networks
  • Layers in Neural Networks

Neuron Models

enter image description here

  • Perceptron
  • Sigmoid Neuron
  • Binary Threshold
  • Rectifier
  • Stochastic Binary

Cost Functions (A Loss or Objective function)

  • Gradient Descent
  • Gradient Boosting
  • Backpropagation
  • Stochastic Gradient Descent

Implementing the classic MNIST dataset problem

  • A Neural Network for handwritten digit recognition
  • Classification using individual pixels

Image Classification

  • A simple implementation using deeper networks enter image description here


  • Expanding the Neural Network using Google's Library for Machine Learning Might change to Caffe - nVIDIA's library for Machine Learning

Deep Learning

  • A brief introduction to Deep Learning practices
  • Auto Encoders
  • Other areas of Deep Learning (A qualitative study)

A brief introduction to Deep Learning practices


User Prerequisites

  • Core Python - lists, dict, string including functions and classes
  • NumPy, SciPy - not necessary but preferred
  • Elementary Calculus - Differentiation and Integration (Understanding qualitatively is enough)
  • Linear Algebra

System Requirements

  • 32/64-bit Windows/Linux architecture with at least 2GB RAM
  • Python3 compiler with NumPy, SciPy and TensorFlow library
  • PDF reader

Other Requirements but not necessarily needed

  • Anaconda3 (or support for ipynb files, Jupyter preferred)
  • A graphic card

Content URLs:


Speaker Info:

Aniket Chowdhury

While I have been programming for more than a decade, my chosen language for the lesser half of the decade has been C++, with a wandering interest in Java, MySql, PHP and Ruby. The last few years were spent in cultivating the language we now all know as Python. The enamoured feasibility of the language over C++ and the ease of understanding over PERL. While being a bit slower due to being it's interpreted nature, better speed benchmarks are being discover by it's PyPy implementation.

My field of interest is Deep Neural Networks. Machine Learning may perhaps helps us to cure even cancer using gene sequencing.

Apart from that I am an avid reader. I read book from all genres and time. My hobbies include football, music, art, drama and of course, programming.

Speaker Links:




Section: Data science
Type: Workshops
Target Audience: Intermediate
Last Updated:

Well written proposal. Exhaustive. Looking forward to this.

Anand B Pillai (~pythonhacker)

The workshop hasn't been yet confirmed. I am still waiting for the invitation. But, I do hope to meet see you there. This is surely going to be a very exciting workshop.

Aniket Chowdhury (~aniket22)

Login to add a new comment.