Stimulating design reuse through automated coding

In designing new product the ability to retrieve drawings of existing components is important if costs are to be controlled by preventing unnecessary duplication if parts. Component coding and classification systems have been used successfully for these purposes but suffer from high operational costs and poor usability arising directly from the manual nature of the coding process itself. A new version of an existing coding system (CAMAC) has been developed to reduce costs by automatically coding engineering drawings. Usability is improved be supporting searches based on a drawing or sketch of the desired component. Test results from a database of several thousand drawings are presented.

Efficient training of RBF networks for classification.

Radial Basis Function networks with linear outputs are often used in regression problems because they can be substantially faster to train than Multi-layer Perceptrons. For classification problems, the use of linear outputs is less appropriate as the outputs are not guaranteed to represent probabilities. We show how RBFs with logistic and softmax outputs can be trained efficiently using the Fisher scoring algorithm. This approach can be used with any model which consists of a generalised linear output function applied to a model which is linear in its parameters. We compare this approach with standard non-linear optimisation algorithms on a number of datasets.

Mean field methods for classification with Gaussian processes

We discuss the Application of TAP mean field methods known from Statistical Mechanics of disordered systems to Bayesian classification with Gaussian processes. In contrast to previous applications, no knowledge about the distribution of inputs is needed. Simulation results for the Sonar data set are given.

Gaussian processes for classification: mean field algorithms:mean-field algorithms

We derive a mean field algorithm for binary classification with Gaussian processes which is based on the TAP approach originally proposed in Statistical Physics of disordered systems. The theory also yields an approximate leave-one-out estimator for the generalization error which is computed with no extra computational cost. We show that from the TAP approach, it is possible to derive both a simpler 'naive' mean field theory and support vector machines (SVM) as limiting cases. For both mean field algorithms and support vectors machines, simulation results for three small benchmark data sets are presented. They show 1. that one may get state of the art performance by using the leave-one-out estimator for model selection and 2. the built-in leave-one-out estimators are extremely precise when compared to the exact leave-one-out estimate. The latter result is a taken as a strong support for the internal consistency of the mean field approach.

Retrieving designs from a sketch using an automated GT coding and classification system

While the retrieval of existing designs to prevent unnecessary duplication of parts is a recognised strategy in the control of design costs the available techniques to achieve this, even in product data management systems, are limited in performance or require large resources. A novel system has been developed based on a new version of an existing coding system (CAMAC) that allows automatic coding of engineering drawings and their subsequent retrieval using a drawing of the desired component as the input. The ability to find designs using a detail drawing rather than textual descriptions is a significant achievement in itself. Previous testing of the system has demonstrated this capability but if a means could be found to find parts from a simple sketch then its practical application would be much more effective. This paper describes the development and testing of such a search capability using a database of over 3000 engineering components.

The effect of flow regimes in the distillation of systems containing two liquid phases

Single phase solutions containing three components have been observed to exhibit foaminess near a single to two liquid phase boundary. It was seen, in a sintered plate column under mass transfer conditions, that distillation systems where the liquid appeared as one phase in one part of a column and two phases in another part, exhibited foaminess when the liquid concentration was near the one phase to two phase boundary. Various ternary systems have been studied in a 50 plate. 30mm i.d. Oldershaw column and it was observed that severe foaming occurred in the middle section of the column near the one liquid phase to two liquid phase boundary and no foaming occurred at the end of the column where liquid was either one phase or two phase. This is known as Ross type foam. Mass transfer experiments with Ross type ternary systems have been carried out in a perspex simulator with small and large hole diameter trays. It was observed that by removal of the more volatile component, Ross type foam did not build up on the tray. Severe entrainment of liquid was observed in all cases leading to a 'dry' tray, even with a low free area small diameter hole tray which was expected to produce a bubbly mixture. Entrainment was more severe for high gas superficial velocities and large hole diameters. This behaviour is quite different from the build up of foam observed when one liquid phase/two liquid phase Ross systems were contacted with air above a small sintered disc or with vapour in an Oldershaw distillation column. This observation explains why distillation columns processing mixtures which change from one liquid phase to two liquid phases (or vice versa) must be severely derated to avoid flooding. Single liquid phase holdups at the spray to bubbly transition were measured using a perspex simulator similar to that of Porter & Wong (17). i.e. with no liquid cross flow. A light transmission technique was used to measure the transition from spray regime to bubbly regime. The effect of tray thickness and the ratio of hole diameter to tray thickness on the transition was evaluated using trays of the same hole diameter and free area but having thickness of 2.38 mm, 4 mm, and 6.35 mm. The liquid holdup at the transition was less with the thin metal trays. This result may be interpreted by the theory of Lockett (101), which predicts the transition liquid holdup in terms of the angle of the gas iet leaving the holes in the sieve plate. All the existing correlations have been compared and none were found to be satisfactory and these correlations have been modified in view of the experimental results obtained. A new correlation has been proposed which takes into account the effect of the hole diameter to tray thickness ratio on the transition and good agreement was obtained between the experimental results and the correlated values of the liquid holdup at the transition. Results have been obtained for two immiscible liquids [kerosene and water] on trays to determine whether foaming can be eliminated by operating in the spray regime. Kerosene was added to a fixed volume of water or water was added to a fixed volume of kerosene. In both cases, there was a transition from spray to bubbly. In the water fixed system. the liquid holdup at the transition was slightly less than the pure kerosene system. Whilst for the kerosene fixed system, the transition occurred at much lower liquid holdups. Trends In the results were similar to those for single liquid phase. New correlations have been proposed for the two cases. It has been found that Ross type foams, observed in a sintered plate column and in the Oldershaw column can be eliminated by either carrying out the separation in a packed column or by the addition of defoaming additives.

Inhibition of microbial colonization of shipboard fuel systems

The aim of the investigation was to study the problem of colonization of shipboard fuel systems and to examine the effect of a number of environmental factors on microbial growth and survival in order to find potential preservative treatments. A variety of microbial species were isolated from samples taken from fuel storage tanks. Bacteria were more numerous than yeasts or fungi and most microorganisms were found at the fuel/water interface. 1he salinity, pH and phosphate concentration of some water bottoms were characteristic of sea water. Others were brackish, acidic and varied in phosphate content. Microorganisms were cultured under a number of environmental conditions. After prolonged incubation, the inoculum size had no effect on the final biomass of Cladosporium resinae but the time required to achieve the final mass decreased with increasing spore number. Undecane supported better growth of the fungus than diesel fuel and of four types of diesel fuel, two allowed more profuse growth. With sea water as the aqueous phase, a number of isolates were inhibited but the addition of nutrients allowed the development of many of the organisms. Agitation increased the growth of C. resinae on glucose but inhibited it on hydrocarbons. The optimum temperature fgr growth of C. resinae on surface culture lay between 25º C and 30º C and growth was evident at 5º C but not at 45º C. In aqueous suspension, 90% of spores were inactivated in around 60 hours at 45ºC and the same proportion of spores of C. resinae and Penicillium corylophilum were destroyed after about 30 seconds at 65ºC. The majority of bacteria and all yeasts in a water bottom sample were killed within 10 seconds at this temperature. An increase in the concentration of an organo-boron compound caused more rapid inactivation of C. resinae spores and raising the temperature from 25ºC to 45°C significantly enhanced the potency of the biocide.

Classification and contextual enhancement of remotely sensed data

The aims of the project were twofold: 1) To investigate classification procedures for remotely sensed digital data, in order to develop modifications to existing algorithms and propose novel classification procedures; and 2) To investigate and develop algorithms for contextual enhancement of classified imagery in order to increase classification accuracy. The following classifiers were examined: box, decision tree, minimum distance, maximum likelihood. In addition to these the following algorithms were developed during the course of the research: deviant distance, look up table and an automated decision tree classifier using expert systems technology. Clustering techniques for unsupervised classification were also investigated. Contextual enhancements investigated were: mode filters, small area replacement and Wharton's CONAN algorithm. Additionally methods for noise and edge based declassification and contextual reclassification, non-probabilitic relaxation and relaxation based on Markov chain theory were developed. The advantages of per-field classifiers and Geographical Information Systems were investigated. The conclusions presented suggest suitable combinations of classifier and contextual enhancement, given user accuracy requirements and time constraints. These were then tested for validity using a different data set. A brief examination of the utility of the recommended contextual algorithms for reducing the effects of data noise was also carried out.

The application of artificial neural networks to the interpretation and classification of freshwater benthic invertebrate communities

This thesis presents a thorough and principled investigation into the application of artificial neural networks to the biological monitoring of freshwater. It contains original ideas on the classification and interpretation of benthic macroinvertebrates, and aims to demonstrate their superiority over the biotic systems currently used in the UK to report river water quality. The conceptual basis of a new biological classification system is described, and a full review and analysis of a number of river data sets is presented. The biological classification is compared to the common biotic systems using data from the Upper Trent catchment. This data contained 292 expertly classified invertebrate samples identified to mixed taxonomic levels. The neural network experimental work concentrates on the classification of the invertebrate samples into biological class, where only a subset of the sample is used to form the classification. Other experimentation is conducted into the identification of novel input samples, the classification of samples from different biotopes and the use of prior information in the neural network models. The biological classification is shown to provide an intuitive interpretation of a graphical representation, generated without reference to the class labels, of the Upper Trent data. The selection of key indicator taxa is considered using three different approaches; one novel, one from information theory and one from classical statistical methods. Good indicators of quality class based on these analyses are found to be in good agreement with those chosen by a domain expert. The change in information associated with different levels of identification and enumeration of taxa is quantified. The feasibility of using neural network classifiers and predictors to develop numeric criteria for the biological assessment of sediment contamination in the Great Lakes is also investigated.

Mass transfer in fluidized bed drying of moist particulate

Bubbling fluidized bed technology is one of the most effective mean for interaction between solid and gas flow, mainly due to its good mixing and high heat and mass transfer rate. It has been widely used at a commercial scale for drying of grains such as in pharmaceutical, fertilizers and food industries. When applied to drying of non-pours moist solid particles, the water is drawn-off driven by the difference in water concentration between the solid phase and the fluidizing gas. In most cases, the fluidizing gas or drying agent is air. Despite of the simplicity of its operation, the design of a bubbling fluidized bed dryer requires an understanding of the combined complexity in hydrodynamics and the mass transfer mechanism. On the other hand, reliable mass transfer coefficient equations are also required to satisfy the growing interest in mathematical modelling and simulation, for accurate prediction of the process kinetics. This chapter presents an overview of the various mechanisms contributing to particulate drying in a bubbling fluidized bed and the mass transfer coefficient corresponding to each mechanism. In addition, a case study on measuring the overall mass transfer coefficient is discussed. These measurements are then used for the validation of mass transfer coefficient correlations and for assessing the various assumptions used in developing these correlations.

Protein-protein interactions classification from text via local learning with class priors

Text classification is essential for narrowing down the number of documents relevant to a particular topic for further pursual, especially when searching through large biomedical databases. Protein-protein interactions are an example of such a topic with databases being devoted specifically to them. This paper proposed a semi-supervised learning algorithm via local learning with class priors (LL-CP) for biomedical text classification where unlabeled data points are classified in a vector space based on their proximity to labeled nodes. The algorithm has been evaluated on a corpus of biomedical documents to identify abstracts containing information about protein-protein interactions with promising results. Experimental results show that LL-CP outperforms the traditional semisupervised learning algorithms such as SVMand it also performs better than local learning without incorporating class priors.

Mass transfer and chemical reaction on a distillation plate

A multistage distillation column in which mass transfer and a reversible chemical reaction occurred simultaneously, has been investigated to formulate a technique by which this process can be analysed or predicted. A transesterification reaction between ethyl alcohol and butyl acetate, catalysed by concentrated sulphuric acid, was selected for the investigation and all the components were analysed on a gas liquid chromatograph. The transesterification reaction kinetics have been studied in a batch reactor for catalyst concentrations of 0.1 - 1.0 weight percent and temperatures between 21.4 and 85.0 °C. The reaction was found to be second order and dependent on the catalyst concentration at a given temperature. The vapour liquid equilibrium data for six binary, four ternary and one quaternary systems are measured at atmospheric pressure using a modified Cathala dynamic equilibrium still. The systems with the exception of ethyl alcohol - butyl alcohol mixtures, were found to be non-ideal. Multicomponent vapour liquid equilibrium compositions were predicted by a computer programme which utilised the Van Laar constants obtained from the binary data sets. Good agreement was obtained between the predicted and experimental quaternary equilibrium vapour compositions. Continuous transesterification experiments were carried out in a six stage sieve plate distillation column. The column was 3" in internal diameter and of unit construction in glass. The plates were 8" apart and had a free area of 7.7%. Both the liquid and vapour streams were analysed. The component conversion was dependent on the boilup rate and the reflux ratio. Because of the presence of the reaction, the concentration of one of the lighter components increased below the feed plate. In the same region a highly developed foam was formed due to the presence of the catalyst. The experimental results were analysed by the solution of a series of simultaneous enthalpy and mass equations. Good agreement was obtained between the experimental and calculated results.

On the hierarchical classification of G protein-coupled receptors

MOTIVATION: G protein-coupled receptors (GPCRs) play an important role in many physiological systems by transducing an extracellular signal into an intracellular response. Over 50% of all marketed drugs are targeted towards a GPCR. There is considerable interest in developing an algorithm that could effectively predict the function of a GPCR from its primary sequence. Such an algorithm is useful not only in identifying novel GPCR sequences but in characterizing the interrelationships between known GPCRs. RESULTS: An alignment-free approach to GPCR classification has been developed using techniques drawn from data mining and proteochemometrics. A dataset of over 8000 sequences was constructed to train the algorithm. This represents one of the largest GPCR datasets currently available. A predictive algorithm was developed based upon the simplest reasonable numerical representation of the protein's physicochemical properties. A selective top-down approach was developed, which used a hierarchical classifier to assign sequences to subdivisions within the GPCR hierarchy. The predictive performance of the algorithm was assessed against several standard data mining classifiers and further validated against Support Vector Machine-based GPCR prediction servers. The selective top-down approach achieves significantly higher accuracy than standard data mining methods in almost all cases.