Opening the Black Box

Neural Networks have fuelled a machine learning revolution over the past decade that has led to machines accomplishing amazingly complex tasks. However, these models are largely black boxes: we know they work, but they are so complex (up to hundreds of millions of parameters!) that we struggle to understand the limits of such systems. Yet understanding networks becomes extremely important as networks are deployed in safety critical fields, like medicine and autonomous vehicles. Models must be vetted for robustness against adversarial examples, biases need to be identified and compensated for, and boundaries for what the network will produce need to be identified.

What MLC@Home Does

MLC@Home provides an open, collaborative platform for researchers studying machine learning comprehension. It allows us to train thousands of networks in parallel, with tightly controlled inputs, hyperparameters, and network structures. We use this to gain insights into these complex models.

MLC@Home's initial project, the Machine Learning Dataset Generator (MLDS), will generate a large dataset of simple networks trained with both clean and adversarial data. To our knowledge, this is the first dataset of its kind. MLC@Home also welcomes project proposals from other researchers aligned with this research area. MLC@Home requests that all data generated by our supported projects be made available to the public, and that, where possible, any papers and analysis be made public as well.

How You Can Help

We ask for volunteers to donate some of their background computing time to help us continue our research. We use the same time-tested BOINC distributed computing infrastructure as SETI@home and Rosetta@home. Follow these steps to join:

• Install BOINC [ download ]
• Open BOINC Manager
• Click "Tools"->"Add Project"
• Select "MLC@Home" from the menu
• Follow the prompts to create an account

You're done! BOINC takes over from there, running in the background doing science when your computer is idle. You can follow your progress on this project by choosing "BOINC Project Page" from the menu at the top of this page.