Pseudo-contractions in belief revision

Belief Revision addresses the problem of how to change epistemic states, usually represented in the literature by sets of logical sentences. Solid theoretical results were consolidated with the AGM paradigm, which deals with theories (logically closed sets of sentences). After that, the theory was extended to belief bases, that is, arbitrary sets of sentences. Besides all this theoretical framework, AI researchers face serious difficulties when trying to implement belief revision systems. One of the major complications is the closure required by AGM theory, which cannot be easily computed. Even belief bases, which do not require closure, seem to be improper for practical purposes, since their changes are usually very rigid (syntax dependent). Some operations, known as pseudo-contractions, are in the middle ground between belief set change and belief base change. In the present work we have proposed a new pseudo-contraction operation, studied its properties and characterized it. We have also found connections between this operator and some other pseudo-contractions.

Hybrid convex combinations for IIR system identification.

The low complexity of IIR adaptive filters (AFs) is specially appealing to realtime applications but some drawbacks have been preventing their widespread use so far. For gradient based IIR AFs, adverse operational conditions cause convergence problems in system identification scenarios: underdamped and clustered poles, undermodelling or non-white input signals lead to error surfaces where the adaptation nearly stops on large plateaus or get stuck at sub-optimal local minima that can not be identified as such a priori. Furthermore, the non-stationarity in the input regressor brought by the filter recursivity and the approximations made by the update rules of the stochastic gradient algorithms constrain the learning step size to small values, causing slow convergence. In this work, we propose IIR performance enhancement strategies based on hybrid combinations of AFs that achieve higher convergence rates than ordinary IIR AFs while keeping the stability.