A novel information theory method for filter feature selection

In this paper, we propose a novel filter for feature selection. Such filter relies on the estimation of the mutual information between features and classes. We bypass the estimation of the probability density function with the aid of the entropic-graphs approximation of Rényi entropy, and the subsequent approximation of the Shannon one. The complexity of such bypassing process does not depend on the number of dimensions but on the number of patterns/samples, and thus the curse of dimensionality is circumvented. We show that it is then possible to outperform a greedy algorithm based on the maximal relevance and minimal redundancy criterion. We successfully test our method both in the contexts of image classification and microarray data classification.

Cotunneling theory of atomic spin inelastic electron tunneling spectroscopy

We propose cotunneling as the microscopic mechanism that makes possible inelastic electron tunneling spectroscopy of magnetic atoms in surfaces for a wide range of systems, including single magnetic adatoms, molecules, and molecular stacks. We describe electronic transport between the scanning tip and the conducting surface through the magnetic system (MS) with a generalized Anderson model, without making use of effective spin models. Transport and spin dynamics are described with an effective cotunneling Hamiltonian in which the correlations in the magnetic system are calculated exactly and the coupling to the electrodes is included up to second order in the tip MS and MS substrate. In the adequate limit our approach is equivalent to the phenomenological Kondo exchange model that successfully describes the experiments. We apply our method to study in detail inelastic transport in two systems, stacks of cobalt phthalocyanines and a single Mn atom on Cu2N. Our method accounts for both the large contribution of the inelastic spin exchange events to the conductance and the observed conductance asymmetry.