‘Minimalist machine learning’ algorithm analyzes complex microscopy and other images from very little data

http://www.kurzweilai.net/minimalist-machine-learning-algorithm-analyzes-complex-microscopy-and-other-images-from-very-little-data

I tend to work with datasets that are fairly small compared to normal AI datasets and it is very difficult and expensive to gain more data. With that in mind I found this to be a really exciting development. Is anyone here familiar with this approach? I feel like I need a little help to fully understand the approach.

1 Like

Like in this article MIT’s Feature Labs Helps Companies Develop Faster Machine Learning Algorithms.

I have a set of Scanning electron Microscope data and want to use some form of image recognition or classification for the hundreds of varieties of minerals. The problem is that depending on researcher you have different image size references or strength of imaging based on machine. On another note, I suppose with XRD data it is easier to use tabular data for peak counts vs diffraction angles to determine minerals.