2 ms·
In practice, this could be really bad for machine learning when applied to cell images (I work for a company that grows modified human cells and collects the im
by dekhn 8y ago
In practice, this could be really bad for machine learning when applied to cell images (I work for a company that grows modified human cells and collects the images and applies DNNs to them as classifier). DMEM (or FBS, which we use) would effectively cause a systemic bias that would be hard to correct for.