denwid / pcml-cheatsheet Goto Github PK

None of my buisness since I am not enrolled in PCML, but your document looks like being in US Letter format.

Add the a4paper option to your geometry package to gain more space (and mess up with your formatting ;) )

As one would say, gl&hf!

Hi,
Since I removed some stuff (figures) and did not wanted to impose you those modifications if you disagree with them, I made a new branch.

• Table of computational complexities for algorithms (partly done)
• Talble of cost functions (partly done)
• Bias-Variance decomposition (partly done), Test error, Train error, k-fold cross validation
• Models: gaussian processes, time-series models, factor models, mixture models, random forest, etc.
• Computational goodness (convex, low computational complexity) vs. Statistical goodness (robust to outliers, ...), No free lunch theorem
• Mathematical Definitions: Efficiency, Optimality, Consistency, Identifiability, Convexity
• SVD, PCA

Maybe:

• Something on Regularization, Model averaging, Dimensionality reduction
• Definitions: Model, Representation, "generalize"