Back to Courses

Introduction to Neural Networks

Learn why neural networks are flexible tools for learning.

Artificial neural networks learn by detecting patterns in huge amounts of information. Much like your own brain, artificial neural nets are flexible, data-processing machines that make predictions and decisions. In fact, the best ones outperform humans at tasks like chess and cancer diagnoses.

In this course, you'll dissect the internal machinery of artificial neural nets through hands-on experimentation, not hairy mathematics. You'll develop intuition about the kinds of problems they are suited to solve, and by the end you’ll be ready to dive into the algorithms, or build one for yourself.



Concepts and

  1. 1


    When traditional AI hit a dead end, artificial neural nets jumped in.

    1. Neural Networks

      Teaching machines to teach themselves

    2. The Computer Vision Problem

      Think image recognition is easy? Try seeing in pixels.

    3. The Folly of Computer Programming

      Why do we need neural networks? Some things just can't be programmed.

    4. Can Computers Learn?

      Do you have to be living to be learning?

  2. 2


    The power of neural networks emerges from these simple building blocks.

    1. Included with
      Brilliant Premium

      The Decision Box

      Meet your first artificial neuron and learn how to encode simple logical operations.

    2. Included with
      Brilliant Premium

      Activation Arithmetic

      You can count on simple artificial neurons — literally.

    3. Included with
      Brilliant Premium

      Decision Boundaries

      Hone your intuition with this graphical model of a binary neuron.

    4. Included with
      Brilliant Premium

      Building an XOR Gate

      Escape the limitations of single neurons by stacking them in layers.

  3. 3


    Connecting neurons together in layers boosts a neural net's performance.

    1. Included with
      Brilliant Premium

      Hidden Layers

      Got some complex data to classify? Try adding a hidden layer to your ANN.

    2. Included with
      Brilliant Premium

      Curve Fitting

      Classifying isn't an ANN's only schtick. They are used to model lots of different data.

    3. Included with
      Brilliant Premium

      Universal Approximator

      Don't think an ANN can model it? Think again — they're universal.

    4. Included with
      Brilliant Premium

      A Shape-Recognizing Network

      Learn how an ANN learns to see — and how you can trick it.