A mechanical model for supervised learning

A broad goal of engineering is to make functional machines with specific, programmed input-output responses. When inputs are specified in advance and few in number, this goal is sought through rational design, changing the system elements to obtain desired responses. In the supervised learning framework of computer science, system parameters (synapses) are modified in response to observed examples of the correct input-output mapping (classification).
In this work, we apply the supervised learning framework to self-folding sheets, using a physically motivated learning rule. The trained sheet classifies labeled forces by folding into discrete folded states. These sheets succeed in classifying real-world data like Iris flowers, and also generalize, similar to other learning algorithms. As learning provides a straightforward framework to programming complex input-output relationships, we hope that implementing these ideas in engineering could usher in new classes of machines, that have so far eluded design.