Projected sequential Gaussian processes: A C++ tool for interpolation of large datasets with heterogeneous noise

Heterogeneous datasets arise naturally in most applications due to the use of a variety of sensors and measuring platforms. Such datasets can be heterogeneous in terms of the error characteristics and sensor models. Treating such data is most naturally accomplished using a Bayesian or model-based geostatistical approach; however, such methods generally scale rather badly with the size of dataset, and require computationally expensive Monte Carlo based inference. Recently within the machine learning and spatial statistics communities many papers have explored the potential of reduced rank representations of the covariance matrix, often referred to as projected or fixed rank approaches. In such methods the covariance function of the posterior process is represented by a reduced rank approximation which is chosen such that there is minimal information loss. In this paper a sequential Bayesian framework for inference in such projected processes is presented. The observations are considered one at a time which avoids the need for high dimensional integrals typically required in a Bayesian approach. A C++ library, gptk, which is part of the INTAMAP web service, is introduced which implements projected, sequential estimation and adds several novel features. In particular the library includes the ability to use a generic observation operator, or sensor model, to permit data fusion. It is also possible to cope with a range of observation error characteristics, including non-Gaussian observation errors. Inference for the covariance parameters is explored, including the impact of the projected process approximation on likelihood profiles. We illustrate the projected sequential method in application to synthetic and real datasets. Limitations and extensions are discussed.     

Sparse on-line gaussian processes

We develop an approach for sparse representations of gaussian process (GP) models (which are Bayesian types of kernel machines) in order to overcome their limitations for large data sets. The method is based on a combination of a Bayesian on-line algorithm, together with a sequential construction of a relevant subsample of the data that fully specifies the prediction of the GP model. By using an appealing parameterization and projection techniques in a reproducing kernel Hilbert space, recursions for the effective parameters and a sparse gaussian approximation of the posterior process are obtained. This allows for both a propagation of predictions and Bayesian error measures. The significance and robustness of our approach are demonstrated on a variety of experiments.     

High-performance diode-pumped Nd:YLF amplifier

We report on the performance of a multipass diode-pumped amplifier design to provide a combination of high gain and efficiency with high stability. A simple rod-cavity design and the establishment of quasi-steady-state operation resulted in a saturated gain of over 6000 at an average output intensity during the pulse train of 7 kW/cm2. The amplifier showed an output stability of 0.2% rms in the short-term and 0.7% rms in the long-term and an output intensity insensitive to input power changes. Zernike analysis of the measurements of pump distortion showed an almost pure astigmatic phase error that can be compensated up to high average power levels.     

Isoelectric focusing and sodium dodecylsulphate electrophoresis of buffer soluble seed proteins of twelve varieties of winged bean (Psophocarpus tetragonolobus L.)

Buffer-soluble seed proteins of 12 varieties of winged bean (Psophocarpus tetragonolobus L.) were separated by ultrathin-layer isoelectric focusing (IEF) and sodium dodecylsulphate electrophoresis(SDS-PAGE). Whereas the molecular weight spectra of the proteins obtained by SDS-PAGE were identical for all varieties studied, the IEF patterns showed variety specificity for each sample. Most of the variety specific protein bands were identified as glycoproteins. The zymograms of the isoelectically focused trypsin inhibitors were variety specific in both activity and isoelectric point distribution.     

A Multimedia, Multilingual Teaching and Training System for Children with Speech Disorders

The development of an audiovisual pronunciation teaching and training method and software system is discussed in this article. The method is designed to help children with speech and hearing disorders gain better control over their speech production. The teaching method is drawn up for progression from individual sound preparation to practice of sounds in sentences for four languages: English, Swedish, Slovenian, and Hungarian. The system is a general language-independent measuring tool and database editor. This database editor makes it possible to construct modules for all participant languages and for different sound groups. Two modules are under development for the system in all languages: one for teaching and training vowels to hearing-impaired children and the other for correction of misarticulated fricative sounds. In the article we present the measuring methods, the used distance score calculations of the visualized speech spectra, and problems in the evaluation of the new multimedia tool.     

The incentive (in)compatibility of group-based qualification systems

ABSTRACT

Tournament organisers supposedly design rules such that a team cannot be strictly better off by exerting a lower effort. However, the European qualification tournaments for recent FIFA soccer World Cups are known to violate this requirement, which inspires our study on the incentive compatibility of similar group-based qualification systems. Theorems listing the sufficient and necessary conditions of strategy-proofness are provided and applied to classify several soccer qualification tournaments for FIFA World Cups and UEFA European Championships. Two reasonable mechanisms are proposed to solve the problem of incentive incompatibility: the first is based on abolishing the anonymity of the matches discarded in the comparison of teams from different groups, while the second involves a rethinking of the seeding procedure. Our results have useful implications for the governing bodies of major sports.     

A chemiluminescent microtiter plate assay for sensitive detection of protein kinase activity

The toxic effects of methamphetamine (METH) (2.5, 5.0 and 10.0 mg/kg) and methylenedioxymethamphetamine (MDMA) (5.0, 10.0 and 20.0 mg/kg) on dopaminergic systems were assessed in the striatum and of the nucleus accumbens in mdr1a wild-type and knockout mice. METH caused significant dose-dependent decreases of dopamine (DA) and DA transporters (DAT) in the striatum and the nucleus accumbens (NAc) of both wild-type and knockout mice. The lowest doses of METH (2.5 mg/kg) caused only small changes in the wild-type, but marked. decreases in the mdr1a knockout mice. The two higher doses (5 mg/kg and 10 mg/kg) caused similar changes in both strains of mice. In contrast to METH, MDMA caused greater percentage decreases in DAT in the wild-type mice. For example, the lowest dose (5 mg/kg) caused significant decreases in DAT in the NAc of wild-type but not of mdr1a knockout mice. The highest dose (20 mg/kg) caused similar changes in both the strains. These results suggest that METH and MDMA interact differentially with P-glycoproteins. These observations document, for the first time, a role for these proteins in the entry of METH and MDMA into the brain via the blood-brain barrier, with P-glycoprotein possibly facilitating the entry of MDMA but interfering with that of METH into the brain.     

The 33-kDa C-terminal domain of Raf-1 protein kinase exhibits a Ras-independent serum- and phorbol ester-induced shift in gel mobility

Experiments were carried out to determine Raf-1 protein kinase domain fragments which exhibit a characteristic electrophoretic mobility shift noted with Raf-1 protein kinase in response to serum and phorbol ester (PMA) treatment of serum-deprived NIH 3T3 cells. Epsilon-epitope tagged 84 kDa Raf-1 holoenzyme (HR-epsilon), as well as the epsilon-epsilon pitope tagged 35 kDa N-terminal (RI-epsilon), 33 kDa mid-portion (RII-epsilon), and 33 kDa C-terminal (RIII-epsilon) fragments of Raf-1 were overexpressed in NIH 3T3 cells. The overexpressed HR-epsilon exhibited a serum- and PMA-induced shift in gel mobility similar to that noted with endogenous Raf-1. The C-terminal RIII-epsilon fragment exhibited a similar shift in gel mobility while the electrophoretic mobility of the N-terminal RI-epsilon fragment remained unchanged. These results suggest that modification(s) within the 33 kDa C-terminal portion of Raf-1 which occur independently of association with Ras may be responsible for the band shift observed with serum and PMA treatment of serum-deprived NIH 3T3 cells.     

Indirect robot model learning for tracking control

Learning task-space tracking control on redundant robot manipulators is an important but difficult problem. A main difficulty is the non-uniqueness of the solution: a task-space trajectory has multiple joint-space trajectories associated, therefore averaging over non-convex solution space needs to be done if treated as a regression problem. A second class of difficulties arise for those robots when the physical model is either too complex or even not available. In this situation machine learning methods may be a suitable alternative to classical approaches. We propose a learning framework for tracking control that is applicable for underactuated or non-rigid robots where an analytical physical model of the robot is unavailable. The proposed framework builds on the insight that tracking problems are well defined in the joint task- and joint-space coordinates and consequently predictions can be obtained via local optimization. Physical experiments show that state-of-the art accuracy can be achieved in both online and offline tracking control learning. Furthermore, we show that the presented method is capable of controlling underactuated robot architectures as well.