Gaussian Process training with input noise

In standard Gaussian Process regression input locations are assumed to be noise free. We present a simple yet effective GP model for training on input points corrupted by i.i.d. Gaussian noise. To make computations tractable we use a local linear expansion about each input point. This allows the input noise to be recast as output noise proportional to the squared gradient of the GP posterior mean. The input noise variances are inferred from the data as extra hyperparameters. They are trained alongside other hyperparameters by the usual method of maximisation of the marginal likelihood. Training uses an iterative scheme, which alternates between optimising the hyperparameters and calculating the posterior gradient. Analytic predictive moments can then be found for Gaussian distributed test points. We compare our model to others over a range of different regression problems and show that it improves over current methods.

Observation of early to full covering stages of ethylene-based CVD of graphene

Scalable growth is essential for graphene-based applications. Recent development has enabled the achievement of the scalability by use of chemical vapor deposition (CVD) at 1000°C with copper as a catalyst and methane as a precursor gas. Here we report our observation of early stage of graphene growth based on an ethylene-based CVD method, capable of reducing the growth temperature to 770°C for monolayer graphene growth on copper. We track the early stages of slow growth under low ethylene flow rate and observe the graphene domain evolution by varying the temperature and growth time. Temperature-dependence of graphene domain density gives an apparent activation energy of 1.0 eV for nucleation.

Factor analysis based VTS discriminative adaptive training

Vector Taylor Series (VTS) model based compensation is a powerful approach for noise robust speech recognition. An important extension to this approach is VTS adaptive training (VAT), which allows canonical models to be estimated on diverse noise-degraded training data. These canonical model can be estimated using EM-based approaches, allowing simple extensions to discriminative VAT (DVAT). However to ensure a diagonal corrupted speech covariance matrix the Jacobian (loading matrix) relating the noise and clean speech is diagonalised. In this work an approach for yielding optimal diagonal loading matrices based on minimising the expected KL-divergence between the diagonal loading matrix and "correct" distributions is proposed. The performance of DVAT using the standard and optimal diagonalisation was evaluated on both in-car collected data and the Aurora4 task. © 2012 IEEE.

N-best error simulation for training spoken dialogue systems

A recent trend in spoken dialogue research is the use of reinforcement learning to train dialogue systems in a simulated environment. Past researchers have shown that the types of errors that are simulated can have a significant effect on simulated dialogue performance. Since modern systems typically receive an N-best list of possible user utterances, it is important to be able to simulate a full N-best list of hypotheses. This paper presents a new method for simulating such errors based on logistic regression, as well as a new method for simulating the structure of N-best lists of semantics and their probabilities, based on the Dirichlet distribution. Off-line evaluations show that the new Dirichlet model results in a much closer match to the receiver operating characteristics (ROC) of the live data. Experiments also show that the logistic model gives confusions that are closer to the type of confusions observed in live situations. The hope is that these new error models will be able to improve the resulting performance of trained dialogue systems. © 2012 IEEE.

Nanobodies raised against monomeric α-synuclein distinguish between fibrils at different maturation stages.

Nanobodies are single-domain fragments of camelid antibodies that are emerging as versatile tools in biotechnology. We describe here the interactions of a specific nanobody, NbSyn87, with the monomeric and fibrillar forms of α-synuclein (αSyn), a 140-residue protein whose aggregation is associated with Parkinson's disease. We have characterized these interactions using a range of biophysical techniques, including nuclear magnetic resonance and circular dichroism spectroscopy, isothermal titration calorimetry and quartz crystal microbalance measurements. In addition, we have compared the results with those that we have reported previously for a different nanobody, NbSyn2, also raised against monomeric αSyn. This comparison indicates that NbSyn87 and NbSyn2 bind with nanomolar affinity to distinctive epitopes within the C-terminal domain of soluble αSyn, comprising approximately amino acids 118-131 and 137-140, respectively. The calorimetric and quartz crystal microbalance data indicate that the epitopes of both nanobodies are still accessible when αSyn converts into its fibrillar structure. The apparent affinities and other thermodynamic parameters defining the binding between the nanobody and the fibrils, however, vary significantly with the length of time that the process of fibril formation has been allowed to progress and with the conditions under which formation occurs, indicating that the environment of the C-terminal domain of αSyn changes as fibril assembly takes place. These results demonstrate that nanobodies are able to target forms of potentially pathogenic aggregates that differ from each other in relatively minor details of their structure, such as those associated with fibril maturation.