Scaling the Indian Buffet Process via Submodular Maximization

Inference for latent feature models is inherently difficult as the inference space grows exponentially with the size of the input data and number of latent features. In this work, we use Kurihara & Welling (2008)'s maximization-expectation framework to perform approximate MAP inference for linear-Gaussian latent feature models with an Indian Buffet Process (IBP) prior. This formulation yields a submodular function of the features that corresponds to a lower bound on the model evidence. By adding a constant to this function, we obtain a nonnegative submodular function that can be maximized via a greedy algorithm that obtains at least a one-third approximation to the optimal solution. Our inference method scales linearly with the size of the input data, and we show the efficacy of our method on the largest datasets currently analyzed using an IBP model.

Measuring the composition of AlGaN layers in GaN based structures grown on 150 mm Si substrates using (2 0 5) reciprocal space maps

A critical element for the successful growth of GaN device layers on Si is accurate control of the AlGaN buffer layers used to manage strain. Here we present a method for measuring the composition of the AlGaN buffer layers in device structures which makes use of a one-dimensional x-ray detector to provide efficient measurement of a reciprocal space map which covers the full compositional range from AlN to GaN. Combining this with a suitable x-ray reflection with low strain sensitivity it is possible to accurately determine the Al fraction of the buffer layers independent of their relaxation state. © 2013 IOP Publishing Ltd.

Mass chains with passive interconnection: Complex iterative maps and scalability

This paper introduces the problem of passive control of a chain of N identical masses in which there is an identical passive connection between neighbouring masses and a similar connection to a movable point. The problem arises in the design of multi-storey buildings which are subjected to earthquake disturbances, but applies in other situations, for example vehicle platoons. The paper will study the scalar transfer functions from the disturbance to a given intermass displacement. It will be shown that these transfer functions can be conveniently represented in the form of complex iterative maps and that these maps provide a method to establish boundedness in N of the H ∞-norm of these transfer functions for certain choices of interconnection impedance. © 2013 IEEE.