Neuron Sandbox is a new browser-based tool for teaching students how
neural nets work by simulating the operation of a linear threshold unit.
The user interface is heavily scaffolded, and there is a set of problems
of increasing difficulty that students can work through.

You can try it here:  https://www.cs.cmu.edu/~dst/NeuronSandbox

Neuron Sandbox is designed to be used by upper elementary school
students, but it includes advanced features that make it useful even for
undergraduates, such as the ability to switch between threshold and bias
display modes, or to accomodate non-binary inputs.

The software includes a simple editing facility that allows teachers to
construct new problems culturally relevant to their students, e.g., by
referencing their school, community, or family traditions.

Neuron Sandbox was developed by Angela Chen, Neel Pawar, and David
Touretzky at Carnegie Mellon University, with funding from the National
Science Foundation under awards DRL-2049029 and IIS-2112633.  We welcome
your feedback on how it can be improved.