This PR adds COWBIRD, a proposal for enabling ad-tech machine learning optimizations.
Here is a snippet of the proposal that summarizes it:
We propose the use of a paradigm in which the browser is responsible
for evaluating tiny ad-tech models stored on-device. Later, the browser
evaluates the gradients of these models and sends them to an aggregation
service. After these gradients are aggregated, they are sent to the
corresponding ad-tech company, where they can be used to privately
improve model accuracy.
In this proposal, the browser acts as a federated learning platform,
allowing ad-tech companies to optimize toward customizable objectives
with customizable models and features.
This PR adds COWBIRD, a proposal for enabling ad-tech machine learning optimizations.
Here is a snippet of the proposal that summarizes it: