NN#12 — Neural Networks Decoded: Concepts Over Code

Author(s): RSD Studio.ai

Originally published on Towards AI.

Visualizing and Understanding CNNs: The Hidden Machinery of Computer VisionCredits: PrettyStock

This member-only story is on us. Upgrade to access all of Medium.

We have seen in previous article, how limitations of ANN in dealing with spatial data i.e images led to conception of CNNs, inspired by visual cortex of human eye. Now, there is a need to visualize how CNNs work in reality. The true fascination lies in understanding how these systems actually work, how they learn, and what they “see.”

This article interprets inner workings of CNN, how it’s mathematics shape intelligence from images and how it is trained to look for right aspects in a picture.

If you have not read the previous article, do give it a read as it forms the foundation for this one.

Limitations of ANNs: Move to Convolutional Neural Networks

pub.towardsai.net

At the heart of every CNN, there is a deceptively simple operation that traditional neural networks simply cannot match: convolution. This mathematical procedure gives CNNs their name — and their extraordinary power.

Convolution is a mathematical operation in which a smaller matrix is multiplied to various parts of a larger one and the resultant matrix is taken as a downsized version of large matrix. Consider the example below:

Here, we have a 7×7 large matrix… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI