Member-only story

Deep Learning: Deep guide for all your matrix dimensions and calculations!

Skrew Everything
From The Scratch
Published in
6 min readAug 24, 2018
Deep Neural Network

Vote for next post or mention what do you want to see next here:

While learning or programming a deep neural network, the most important thing is to arrange data in the form of matrix.

And using Keras is as easy as arranging the data in the form of matrix and feed it. There is no need to define weights and biases.

But this is From The Scratch! We just go 1 step deep

This is a guide to those who want to use numpy or low-level TensorFlow to code deep models. Just like me 🤙🏻

This post describes:

  • How many ways we can arrange our input data.
  • Different ways to multiply inputs and weights.
  • How to define Weights and Biases based on input data arrangment.
  • How to get the dimensions right.
  • Generalizing the above dimensions problem to calculate the dimensions of Weight and Bias based on Layer.

--

--

From The Scratch
From The Scratch

Published in From The Scratch

Everything From Scratch related to programming for people who are curious to know how things work from bottom.

Skrew Everything
Skrew Everything

Written by Skrew Everything

A wannabe artist 👨‍🎨, but can’t draw 😫. A wannabe athlete 🏃‍♂️,but can’t run 🥵.Found my peace with coding 👨‍💻 and writing ✍️. Twitter.com/SkrewEverything

Responses (4)