A variational perspective on noise-robust speech recognition

Model compensation methods for noise-robust speech recognition have shown good performance. Predictive linear transformations can approximate these methods to balance computational complexity and compensation accuracy. This paper examines both of these approaches from a variational perspective. Using a matched-pair approximation at the component level yields a number of standard forms of model compensation and predictive linear transformations. However, a tighter bound can be obtained by using variational approximations at the state level. Both model-based and predictive linear transform schemes can be implemented in this framework. Preliminary results show that the tighter bound obtained from the state-level variational approach can yield improved performance over standard schemes. © 2011 IEEE.

Two problems with variational expectation maximisation for time-series models

Variational methods are a key component of the approximate inference and learning toolbox. These methods fill an important middle ground, retaining distributional information about uncertainty in latent variables, unlike maximum a posteriori methods (MAP), and yet generally requiring less computational time than Monte Carlo Markov Chain methods. In particular the variational Expectation Maximisation (vEM) and variational Bayes algorithms, both involving variational optimisation of a free-energy, are widely used in time-series modelling. Here, we investigate the success of vEM in simple probabilistic time-series models. First we consider the inference step of vEM, and show that a consequence of the well-known compactness property of variational inference is a failure to propagate uncertainty in time, thus limiting the usefulness of the retained distributional information. In particular, the uncertainty may appear to be smallest precisely when the approximation is poorest. Second, we consider parameter learning and analytically reveal systematic biases in the parameters found by vEM. Surprisingly, simpler variational approximations (such a mean-field) can lead to less bias than more complicated structured approximations.

α-Synuclein senses lipid packing defects and induces lateral expansion of lipids leading to membrane remodeling.

There is increasing evidence for the involvement of lipid membranes in both the functional and pathological properties of α-synuclein (α-Syn). Despite many investigations to characterize the binding of α-Syn to membranes, there is still a lack of understanding of the binding mode linking the properties of lipid membranes to α-Syn insertion into these dynamic structures. Using a combination of an optical biosensing technique and in situ atomic force microscopy, we show that the binding strength of α-Syn is related to the specificity of the lipid environment (the lipid chemistry and steric properties within a bilayer structure) and to the ability of the membranes to accommodate and remodel upon the interaction of α-Syn with lipid membranes. We show that this interaction results in the insertion of α-Syn into the region of the headgroups, inducing a lateral expansion of lipid molecules that can progress to further bilayer remodeling, such as membrane thinning and expansion of lipids out of the membrane plane. We provide new insights into the affinity of α-Syn for lipid packing defects found in vesicles of high curvature and in planar membranes with cone-shaped lipids and suggest a comprehensive model of the interaction between α-Syn and lipid bilayers. The ability of α-Syn to sense lipid packing defects and to remodel membrane structure supports its proposed role in vesicle trafficking.

Investigation of the Young's modulus and thermal expansion of amorphous titania-doped tantala films

The current generation of advanced gravitational wave detectors utilize titania-doped tantala/silica multilayer stacks for their mirror coatings. The properties of the low-refractive-index silica are well known; however, in the absence of detailed direct measurements, the material parameters of Young's modulus and coefficient of thermal expansion (CTE) of the high refractive index material, titania-doped tantala, have been assumed to be equal to values measured for pure tantala coatings. In order to ascertain the true values necessary for thermal noise calculations, we have undertaken measurements of Young's modulus and CTE through the use of nanoindentation and thermal-bending measurements. The measurements were designed to assess the effects of titania doping concentration and post-deposition heat-treatment on the measured values in order to evaluate the possibility of optimizing material parameters to further improve thermal noise in the detector. Young's modulus measurements on pure tantala and 25% and 55% titania-doped tantala show a wide range of values, from 132 to 177 GPa, dependent on both titania concentration and heat-treatment. Measurements of CTE give values of (3.9 +/- 0.1) x 10^-6 K^-1 and (4.9 +/- 0.3) x 10^-6 K^-1 for 25% and 55% titania-doped tantala, respectively, without dependence on post-deposition heat-treatment.

Reactive Many-Body Expansion for a Protonated Water Cluster.

We generalize the standard many-body expansion technique that is used to approximate the total energy of a molecular system to enable the treatment of chemical reactions by quantum chemical techniques. By considering all possible assignments of atoms to monomer units of the many-body expansion and associating suitable weights with each, we construct a potential energy surface that is a smooth function of the nuclear positions. We derive expressions for this reactive many-body expansion energy and describe an algorithm for its evaluation, which scales polynomially with system size, and therefore will make the method feasible for future condensed phase simulations. We demonstrate the accuracy and smoothness of the resulting potential energy surface on a molecular dynamics trajectory of the protonated water hexamer, using the Hartree-Fock method for the many-body term and Møller-Plesset theory for the low order terms of the many-body expansion.