Data-driven advice for ML in Biomedical Informatics

The increased accessibility of ML creates a “choice overload” for biomedical informaticians when it comes to choosing appropriate ML methods and parameters for their datasets. This week at PSB, I’ll be presenting our recent work to tackle this problem with data-driven findings. We ran a large experimental study of several open-source ML algorithms on many open-source datasets to generate recommendations for where to begin. Check out the paper here.