Johannes Soeding

[Johannes Soeding @JohannesSoeding]

@mikhailspivakov @steinaerts We used 4-fold X-validation throughout, so overtraining would cost performance. Also, large k worked just fine.

[Johannes Soeding @JohannesSoeding]

@JohannesSoeding @biorxivpreprint This is very significant considering that the p-value of 1/16 is obtained on 97% of 464 data sets.