We examine gradient descent on unregularised logistic regression problems, with homogeneous linear predictors on linearly separable datasets. We show the predictor converges to the direction of the max-margin (hard margin SVM) solution.

I will complete a couple of proofs from last week's student seminar on determinantal complexity. I will recall the required background for the proofs. On the way, we will learn a cute linear-algebraic fact.

We show how to combine techniques from formal methods and learning for online computation of a strategy that aims at optimizing the expected long-term reward in large systems modelled as Markov decision processes.

In the composite hypothesis testing setting, the detector receives n i.i.d. samples either from a distribution  p∈P or from a distribution q∈Q. It then decides the correct set from which the samples were drawn.

We formulate a general cross validation framework for signal denoising. The general framework is then applied to nonparametric regression methods such as Trend Filtering and Dyadic CART.