On Discriminative Bayesian Network Classifiers and Logistic Regression

Discriminative learning of the parameters in the naive Bayes model is known to be equivalent to a logistic regression problem. Here we show that the same fact holds for much more general Bayesian network models, as long as the corresponding network structure satisfies a certain graph-theoretic property. The property holds for naive Bayes but also for more complex structures such as tree-augmented naive Bayes (TAN) as well as for mixed diagnostic-discriminative structures. Our results imply that for networks satisfying our property, the conditional likelihood cannot have local maxima so that the global maximum can be found by simple local optimization methods. We also show that if this property does not hold, then in general the conditional likelihood can have local, non-global maxima. We illustrate our theoretical results by empirical experiments with local optimization in a conditional naive Bayes model. Furthermore, we provide a heuristic strategy for pruning the number of parameters and relevant features in such models. For many data sets, we obtain good results with heavily pruned submodels containing many fewer parameters than the original naive Bayes model.Editors: Pedro Larrañaga, Jose A. Lozano, Jose M. Peña and Iñaki Inza     

Halogen bonding based recognition processes: a world parallel to hydrogen bonding

Halogen bonding is the noncovalent interaction between halogen atoms (Lewis acids) and neutral or anionic Lewis bases. The main features of the interaction are given, and the close similarity with the hydrogen bonding will become apparent. Some heuristic principles are presented to develop a rational crystal engineering based on halogen bonding. The focus is on halogen-bonded supramolecular architectures given by halocarbons. The potential of the interaction is shown by useful applications in the field of synthetic chemistry, material science, and bioorganic chemistry.     

Structural mean models for compliance analysis in randomized clinical trials and the impact of errors on measures of exposure

Partial compliance with assigned treatment regimes is common in drug trials and calls for a causal analysis of the effect of treatment actually received. As such observed exposure is no longer randomized, selection bias must be carefully accounted for. The framework of potential outcomes allows this by defining a subject-specific treatment-free reference outcome, which may be latent and is modelled in relation to the observed (treated) data. Causal parameters enter these structural models explicitly. In this paper we review recent progress in randomization-based inference for structural mean modelling, from the additive linear model to the structural generalized linear models. An arsenal of tools currently available for standard association regression has steadily been developed in the structural setting, providing many parallel features to help randomization-based inference. We argue that measurement error on exposure is an important practical complication that has, however, not yet been addressed. We show how standard additive linear structural mean models are robust against unbiased measurement error and how efficient, asymptotically unbiased inference can be drawn when the degree of measurement error bias is known. The impact of measurement error is illustrated in a blood pressure example and finite sample properties are verified by simulation. We end with a plea for more and careful use of this methodology and point to directions for further development.     

Process control of lightly salted wild and farmed Atlantic cod (Gadus morhua) by brine injection, brining, and freezing--a low field NMR study

The aim of this study was to use low-field nuclear magnetic resonance (NMR) and traditional chemical methods to investigate the physical and chemical differences in wild and farmed cod processed pre- and postrigor, and how these properties were affected by brine injection, brining, and freezing. In prerigor processed farmed or wild cod, brine injections followed by brining for 2 d, with brine concentrations up to 5.5% and 4%, respectively, were not sufficient to reach a muscle salt concentration of 2% as aimed for, while wild cod processed postrigor had sufficient salt uptake after the same processing. Low-field NMR gave valuable information about the differences in the muscle structure between wild and farmed cod as well as the state of the water in the muscle during brine injection, brining, and during rigor tension. Low-field NMR is, therefore, a valuable tool that can be used to optimize the salting and storing processes of lightly salted cod products from both wild and farmed cod. For farmed cod to be used in the production of lightly salted products further research is needed. Practical Application: Optimal processing of lightly salted cod products is important to the fish industry, due to an increasing market for this product in southern Europe. Farmed cod, which is seen as a potential steady raw material source for this production, differs considerably from its wild counterparts by having other chemical and physical muscle properties, such as lower water content and lower pH. With the processing procedures used today the farmed cod can, therefore, only be used in some of the products, where wild cod is currently used as raw material. It is, therefore, important that the processing of these products is optimized with regard to these differences in the raw material. This study gives a valuable contribution to further studies about optimal combinations of brine injections, brining, and freezing of pre- and postrigor processed farmed compared to wild cod.     

Research-by-design framework for integrating education and research in an intercultural parallel design studio

Intercultural student exchange can positively impact education, research, and society. Research-by-design has been put forward in design education as an approach to explicitly address the integration of education and research. The problem is that design schools develop their own interpretation of this approach, which brings difficulty in virtually comparing and learning from the generated design proposals. This case is evident in intercultural parallel design studios that deal with diverging socio-cultural and institutional traditions. In addressing this problem and maximizing the benefits of intercultural exchange, this study proposes to adopt a meta perspective and use the design as co-evolution model as a framework for analyzing and comparing design data generated by different research-by-design approaches. The study applies this meta perspective on an intercultural architectural design studio jointly organized by Ton Duc Thang University (Vietnam) and Hasselt University (Belgium).     

Structural studies on Ca3Al4MgO10 (C3A2M)—A ternary phase in the system CaO–Al2O3–MgO

In a sequence of temperature-dependent solid-state reactions in the system CaO–Al₂O₃–MgO the formation of the ternary phase Ca₃Al₄MgO₁₀ or C₃A₂M has been studied. Whereas the compound could not be prepared at 1200°C, a yield of 85 wt.-% of Ca₃Al₄MgO₁₀ was obtained at 1320°C (incongruent melting point: 1330°C). Powder diffraction data compare well with results of previous investigations from the 1960s. Single crystals of Ca₃Al₄MgO₁₀ could be retrieved from the sinter-pellets. Basic crystallographic data are as follows: orthorhombic symmetry, space group Pbcm, a = 5.14073(8), b = 16.7576(2), c = 10.70977(16) Å, V = 922.61(2) ų, Z = 4. Using synchrotron diffraction data it was possible to solve the crystal structure. Least-squares refinements resulted in a residual of R(|F|) = 0.021 for 1000 independent observed reflections with I > 2σ(I) and 97 parameters. The structure contains [TO₄]-tetrahedra (T=Al,Mg) forming a three-dimensional (3-D) framework whose topological characteristics have been determined. Al-Mg distributions on the different T-sites have been studied. The calcium cations are located in voids of the network. More than 50 years after its first observation our investigation clarifies the crystal structure of a compound belonging to a system that is of relevance for several fields of materials science.     

Spurenstoffabtrennung mit keramischen Nanofiltrationsmembranen als Rotationsscheibenfilter

Reducing micropollutant pollution of water bodies is an important objective of water management and an integral part of environmental policy. Ceramic nanofiltration membranes were developed as multichannel membranes of increased membrane area and rotating disk filters. The membranes developed show retention of over 80 % for PEG 400. The membranes are currently being tested for the separation of micropollutants from wastewater contaminated with pharmaceuticals. With the help of a downstream oxidative process, the trace substances remaining in the permeate are degraded.     

Fabrication of Ti3SiC2-based pastes for screen printing on paper-derived Al2O3 substrates

This paper describes the development and fabrication of pastes suitable for screen printing process using Ti₃SiC₂ as the ceramic filler and ethyl cellulose as the binder. With the aim of obtaining high quality screen printed films, the influence of different amounts of Ti₃SiC₂ filler (20–40?vol%) and binder (0–5?vol%) on the rheological properties of the pastes was investigated. Samples with higher viscosity, such as pastes containing 30?vol% and 40?vol% Ti₃SiC₂ filler, regardless of the amount of ethyl cellulose, showed a higher printing quality compared to the samples with other compositions. The different paste compositions were screen printed onto paper-derived Al₂O₃ substrates containing 28.6 ± 4.8% open porosity and sintered for 1?h under an argon atmosphere at 1600?°C. X-ray diffraction (XRD) measurements and scanning electron microscopy (SEM) analysis showed that the sintered films contained TiC as a primary phase and Ti₃SiC₂ as a secondary phase. The partial decomposition of Ti₃SiC₂ after sintering can be attributed to residual carbon from the organic additives, which decreases the thermal stability of this material.