1) Statistics as a Bayesian, e.g., Graphical Models
2) Statistics as a Frequentist, e.g., Statistical Learning Theory.
3) Information Theory, e.g., Entropy, MDL as a formalization of Occam's Razor.
4) Other knowledge from Probability and Stochastic Processes, e.g., random walk
5) In addition, Optimization is fairly useful, especially the Convex Optimization.