I still think that #machinelearning is a misnomer. We could have called it "regression machines", "complex regression", "complex pattern generator". That eould have meant less grant money and less Nature papers, but also a lot less confusion about what #ml can and cannot do.