OWL2Vec*: embedding of OWL ontologies

Machine Learning - Semantic embedding of knowledge graphs has been widely studied and used for prediction and statistical analysis tasks across various domains such as Natural Language Processing...     

Superparamagnetic particle dynamics and mixing in a rotating capillary tube with a stationary magnetic field

The dynamics of superparamagnetic particles subject to competing magnetic and viscous drag forces have been examined with a uniform, stationary, external magnetic field. In this approach, competing drag and magnetic forces were created in a fluid suspension of superparamagnetic particles that was confined in a capillary tube; competing viscous drag and magnetic forces were established by rotating the tube. A critical Mason number was determined for conditions under which the rotation of the capillary prevents the formation of chains from individual particles. The statistics of chain length was investigated by image analysis while varying parameters such as the rotation speed and the viscosity of the liquid. The measurements showed that the rate of particle chain formation was decreased with increased viscosity and rotation speed; the particle dynamics could be quantified by the same dimensionless Mason number that has been demonstrated for rotating magnetic fields. The potential for enhancement of mixing in a bioassay was assessed using a fast chemical reaction that was diffusion-limited. Reducing the Mason number below the critical value, so that chains were formed in the fluid, gave rise to a modest improvement in the time to completion of the reaction.     

A comparative study of comminution in rotary and vibratory ball mills

A comparative study of the comminution in rotary and vibratory ball mills using sodium chloride is presented. The many advantages of vibratory over rotary ball milling are outlined and it is shown experimentally that the rate and extent of grinding is higher for the vibratory ball mill. It is further shown that fine particles re-aggregate in the rotary ball mill and ‘cake’ irreversibly. This phenomenon, exclusive to the rotary ball mill, is explained in terms of consolidation and high stress relaxation of the crystals occurring during prolonged grinding.     

Regressing Local to Global Shape Properties for Online Segmentation and Tracking

We propose a novel regression based framework that uses online learned shape information to reconstruct occluded object contours. Our key insight is to regress the global, coarse, properties of shape from its local properties, i.e. its details. We do this by representing shapes using their 2D discrete cosine transforms and by regressing low frequency from high frequency harmonics. We learn this regression model using Locally Weighted Projection Regression which expedites online, incremental learning. After sufficient observation of a set of unoccluded shapes, the learned model can detect occlusion and recover the full shapes from the occluded ones. We demonstrate the ideas using a level-set based tracking system that provides shape and pose, however, the framework could be embedded in any segmentation-based tracking system. Our experiments demonstrate the efficacy of the method on a variety of objects using both real data and artificial data.     

N.m.r. studies on the effect of water on the glass transition of polystyrene

These investigations are concerned with water-polymer interactions in polymer latices. It is known that water can act as a plasticizer for many solid polymers and cause a reduction in the glass transition temperature, T_g, of the amorphous regions. Experiments were carried out to determine whether pulsed n.m.r. techniques could be used to study the T_g of a polymer suspension and hence the influence of water and electrolyte on it. From T₁ and T₂ proton relaxation measurements as a function of temperature on polystyrene latex systems it was shown that the presence of water lowers the T_g of the polymer particles (by about 10°C), the effect being slightly greater in the presence of concentrated electrolyte. The extent of electrolyte penetration into the particles was deduced by studying relaxation as a function of particle diameter in latices containing paramagnetic Mn²⁺ ions. Using simple theories of relaxation and spin diffusion it was concluded that for all but the smallest particles electrolyte penetration is restricted to a very thin shell of the order of 1 nm. These conclusions were supported by the results of similar measurements on PTFE particles.     

The Instabilisation of Dye-Fibre Bonds by Peroxide Treatments

The results are presented of an investigation into the instability induced in dye-fibre bonds by peroxide treatments. The nature of the reactions involved are discussed.     

The mechanism of photoprotection of wool by UV absorbers of the 2-hydroxybenzophenone class

The mechanism by which sulfonated 2-hydroxybenzophenone ultraviolet (UV) absorbers protect wool fabric against photoyellowing has been studied. The absorption and luminescence properties of two 2-hydroxybenzophenone-5-sulfonates and four 2-hydroxybenzophenone-2′hyphen;sulfonates are compared. The 2-hydroxy-benzophenones are nonfluorescent in wool, but they all show phosphorescence at room temperature. The phosphorescence is attributed to the phenolate anion produced by ionization of the hydroxyl group. The nature of substituents in the 3-position and 4-position of the phenolic ring is important in controlling the formation of luminescent species which may initiate unwanted photochemical reactions in wool. The protective screening effect of each absorber in wool fabric has been determined theoretically, using a method of analysis based on diffuse reflectance spectroscopy; the results are compared with the values determined experimentally by measuring the changes in yellowness that result when treated and untreated fabrics are exposed to fluorescent sunlamps (maximum energy at 310 nm). In all cases, UV screening was shown to be the important mode of photostabilization. The levels of photoprotection were much higher than those reported previously for treated fabrics that were exposed for long periods to simulated sunlight, behind glass. Possible reasons for the poor long-term performance of the 2-hydroxybenzophenones in wool are discussed.     

How to “alternatize” a clustering algorithm

Given a clustering algorithm, how can we adapt it to find multiple, nonredundant, high-quality clusterings? We focus on algorithms based on vector quantization and describe a framework for automatic ‘alternatization’ of such algorithms. Our framework works in both simultaneous and sequential learning formulations and can mine an arbitrary number of alternative clusterings. We demonstrate its applicability to various clustering algorithms—k-means, spectral clustering, constrained clustering, and co-clustering—and effectiveness in mining a variety of datasets.     

Cyclic voltammetry on electrode surfaces covered with porous layers: An analysis of electron transfer kinetics at single-walled carbon nanotube modified electrodes

Cyclic voltammetry is recorded of the oxidation of ferrocyanide on a glassy carbon electrode modified with multiple layers of single-walled carbon nanotubes. The current response is interpreted in terms of semi-infinite planar diffusion towards the macro-electrode surface and in terms of oxidation of the electroactive species trapped in pockets in between the nanotubes. A thin layer model is used to illustrate the effects of diffusion within a porous layer. It is found that a semi-infinite planar diffusion model alone is not appropriate for interpreting the kinetics of the electron transfer at this electrode surface. In particular, caution should be exercised in respect of comparing voltammetric peak-to-peak potential separations between naked electrodes and nanotube-modified electrodes for the inference of electrocatalysis via electron transfer via the nanotubes.     

Minimal ordering constraints for some families of variable symmetries

Variable symmetries in a constraint satisfaction problem can be broken by adding lexicographic ordering constraints. Existing general methods of generating such sets of ordering constraints can require a huge number of constraints. This adds an unacceptable overhead to the solving process. Methods exist by which this large set of ordering constraints can be reduced to a much smaller set automatically, but their application is also prohibitively costly. In contrast, this paper takes a bottom-up approach. It examines some commonly-occurring families of groups and derives a minimal set of ordering constraints sufficient to break the symmetry each group describes. These minimal sets are then used as building blocks to generate minimal sets of ordering constraints for groups constructed via direct and imprimitive wreath products. Experimental results confirm the value of minimal sets of ordering constraints, which can now be generated much more cheaply than with previous methods.