Best of CASE District VII Award
 

Institution: California Institute of Technology

Title of entry: Caltech Science Exchange: How Do Computers Learn?

About this entry: Illustration was created for the Caltech Science Exchange, a web resource dedicated to clear and credible explanations of high-profile science and engineering topics. Machine learning is a process through which computerized systems use human-supplied data and feedback to independently make decisions and predictions, typically becoming more accurate with continual training. This contrasts with traditional computing, in which every action taken by a computer must be pre-programmed.

Computer scientists choose different machine learning approaches depending how the system will be used. For example, reinforcement learning teaches a system as it interacts with an environment by offering it rewards when it performs an action correctly. Two other common approaches that use data are supervised learning, which applies to the computer-vision systems used in autonomous vehicles, and unsupervised learning, which is used when data need to be clustered (for example, audience segmentation for streaming services or product recommendations to online shoppers).

This comic illustrates supervised and unsupervised approaches to machine learning.