Source: Deep Learning on Medium
Interactive Visualization System that Helps Students Better Understand and Learn CNNs
This research summary is just one of many that are distributed weekly on the AI scholar newsletter. To start receiving the weekly newsletter, sign up here.
Artificial intelligence (AI) has grown tremendously in just a few years ushering us into the AI era. We now have self-driving cars, contemporary chatbots, high-end robots, recommender systems, advanced diagnostics systems, and more.
Almost every research field is now using AI. And this AI re-birth is being driven by deep learning’s power and flexibility.
A branch of machine learning, deep learning which was inspired by the human brain structure has experienced a tremendous recent research boost and is solving numerous problems. The technique delivers state of the art results in many applications including computer vision, natural language processing, bioinformatics, among others.
However, there are numerous types of network layers with different structures and underlying mathematical operations. These networks usually leverage many layers of operations to reach a final decision. That said, understanding deep learning models necessitates that users keep track of both low-level and high-level mathematical operations and integration within a network.
Convolutional Neural Networks (CNNs) 101
Deep learning advances specifically the fact that its implementation has helped solve previously hard problems has attracted and inspired both experts and non-experts. However, due to the complexity of deep learning models, it is difficult for learners to take the first baby steps towards this exciting technology.
To address the challenge of complexity in deep learning models, researchers have developed CNN 101, an interactive visualization system that helps students better understand and learn CNNs, which is a foundational deep learning model architecture.
Built using modern web technologies, CNN 101 runs locally in a users’ web browser without the need for specialized hardware, broadening the public’s education access to deep learning techniques. Through tightly integrated interactive views, CNN 101 offers both overview and detailed descriptions of how a model works by explaining convolution, activation, and pooling operations at a single-neuron-level as well as layer level.
As such, it joins the already existing research work that aims to explain the complicated machine learning algorithms with interactive visualization.
Potential Uses and Effects
CNN 101 is another step towards making deep learning more available as well as a step towards building design principles for future deep learning educational tools. It applies interactive visualizing techniques providing users with an easier way to learn deep learning mechanisms and build up neural network intuitions.
The authors plan to extend CNN 101’s capabilities to support further user customization and personalized learning. They are also working to deploy and open-source CNN 101, similar to TensorFlow Playground and GAN Lab so that it will be easily accessible by learners from all over the world.
Watch CNN 101 Demo Video Here
Read more: CNN 101