Learning Binary Code Representations for Effective and Efficient Image Retrieval

The size of online image datasets is constantly increasing. Considering an image dataset with millions of images, image retrieval becomes a seemingly intractable problem for exhaustive similarity search algorithms. Hashing methods, which encodes high-dimensional descriptors into compact binary strings, have become very popular because of their high efficiency in search and storage capacity. In the first part, we propose a multimodal retrieval method based on latent feature models. The procedure consists of a nonparametric Bayesian framework for learning underlying semantically meaningful abstract features in a multimodal dataset, a probabilistic retrieval model that allows cross-modal queries and an extension model for relevance feedback. In the second part, we focus on supervised hashing with kernels. We describe a flexible hashing procedure that treats binary codes and pairwise semantic similarity as latent and observed variables, respectively, in a probabilistic model based on Gaussian processes for binary classification. We present a scalable inference algorithm with the sparse pseudo-input Gaussian process (SPGP) model and distributed computing. In the last part, we define an incremental hashing strategy for dynamic databases where new images are added to the databases frequently. The method is based on a two-stage classification framework using binary and multi-class SVMs. The proposed method also enforces balance in binary codes by an imbalance penalty to obtain higher quality binary codes. We learn hash functions by an efficient algorithm where the NP-hard problem of finding optimal binary codes is solved via cyclic coordinate descent and SVMs are trained in a parallelized incremental manner. For modifications like adding images from an unseen class, we propose an incremental procedure for effective and efficient updates to the previous hash functions. Experiments on three large-scale image datasets demonstrate that the incremental strategy is capable of efficiently updating hash functions to the same retrieval performance as hashing from scratch.

Calculation of sample size for stroke trials assessing functional outcome: comparison of binary and ordinal approaches

Background Many acute stroke trials have given neutral results. Sub-optimal statistical analyses may be failing to detect efficacy. Methods which take account of the ordinal nature of functional outcome data are more efficient. We compare sample size calculations for dichotomous and ordinal outcomes for use in stroke trials. Methods Data from stroke trials studying the effects of interventions known to positively or negatively alter functional outcome – Rankin Scale and Barthel Index – were assessed. Sample size was calculated using comparisons of proportions, means, medians (according to Payne), and ordinal data (according to Whitehead). The sample sizes gained from each method were compared using Friedman 2 way ANOVA. Results Fifty-five comparisons (54 173 patients) of active vs. control treatment were assessed. Estimated sample sizes differed significantly depending on the method of calculation (Po00001). The ordering of the methods showed that the ordinal method of Whitehead and comparison of means produced significantly lower sample sizes than the other methods. The ordinal data method on average reduced sample size by 28% (inter-quartile range 14–53%) compared with the comparison of proportions; however, a 22% increase in sample size was seen with the ordinal method for trials assessing thrombolysis. The comparison of medians method of Payne gave the largest sample sizes. Conclusions Choosing an ordinal rather than binary method of analysis allows most trials to be, on average, smaller by approximately 28% for a given statistical power. Smaller trial sample sizes may help by reducing time to completion, complexity, and financial expense. However, ordinal methods may not be optimal for interventions which both improve functional outcome

Use of ordinal outcomes in vascular prevention trials: comparison with binary outcomes in published trials

Background and Purpose—Vascular prevention trials mostly count “yes/no” (binary) outcome events, eg, stroke/no stroke. Analysis of ordered categorical vascular events (eg, fatal stroke/nonfatal stroke/no stroke) is clinically relevant and could be more powerful statistically. Although this is not a novel idea in the statistical community, ordinal outcomes have not been applied to stroke prevention trials in the past. Methods—Summary data on stroke, myocardial infarction, combined vascular events, and bleeding were obtained by treatment group from published vascular prevention trials. Data were analyzed using 10 statistical approaches which allow comparison of 2 ordinal or binary treatment groups. The results for each statistical test for each trial were then compared using Friedman 2-way analysis of variance with multiple comparison procedures. Results—Across 85 trials (335 305 subjects) the test results differed substantially so that approaches which used the ordinal nature of stroke events (fatal/nonfatal/no stroke) were more efficient than those which combined the data to form 2 groups (P0.0001). The most efficient tests were bootstrapping the difference in mean rank, Mann–Whitney U test, and ordinal logistic regression; 4- and 5-level data were more efficient still. Similar findings were obtained for myocardial infarction, combined vascular outcomes, and bleeding. The findings were consistent across different types, designs and sizes of trial, and for the different types of intervention. Conclusions—When analyzing vascular events from prevention trials, statistical tests which use ordered categorical data are more efficient and are more likely to yield reliable results than binary tests. This approach gives additional information on treatment effects by severity of event and will allow trials to be smaller. (Stroke. 2008;39:000-000.)

Vapor–liquid equilibrium of binary mixtures containing isopropyl acetate and alkanols at 101.32 kPa

Isobaric vapor-liquid equilibria of binary mixtures of isopropyl acetate plus an alkanol (1-propanol, 2-propanol, 1-butanol, or 2-butanol) were measured at 101.32 kPa, using a dynamic recirculating still. An azeotropic behavior was observed only in the mixtures of isopropyl acetate + 2-propanol and isopropyl acetate + 1-propanol. The application of four thermodynamic consistency tests (the Herington test, the Van Ness test, the infinite dilution test, and the pure component test) showed the high quality of the experimental data. Finally, both NRTL and UNIQUAC activity coefficient models were successfully applied in the correlation of the measured data, with the average absolute deviations in vapor phase composition and temperature of 0.01 and 0.16 K, respectively.

Estudo da incorporação de cascalho proveniente da perfuração de poços de petróleo em formulações para tijolos de solo-cimento

The present work is to study the characteristics and technological properties of soil-cement bricks made from binary and ternary mixtures of Portland cement, sand, water, with or without addition of gravel from the drilling of oil wells, which could be used by industry, aiming to improve its performance and reduce cost by using the residue and, consequently, increasing its useful life. The soil-cement bricks are one of the alternatives to masonry construction. These elements, after a short curing period, provide compressive strength similar to that of solid bricks and ceramic blocks, and the higher the resistance the higher the amount of cement used. We used the soil from the city of São José do Mipibu / RN, the banks of the River Baldun, cement CPIIZ-32 and residue of drill cuttings from oil wells drilling onshore wells in the town of Mossley, RN, provided Petrobras. To determine the optimum mix, we studied the inclusion of different residues (100%, 80%, 70%, 60% and 50%) where 15 bodies were made of the test piece. The assessment was made of bricks made from simple compression tests, mass loss by immersion and water absorption. The experimental results proved the efficiency and high utilization of the waste from the drilling of oil wells, making the brick-cement-soil residue with a higher strength and lower water absorption. The best result in terms of mechanical strength and water absorption for the ternary mixture was 10% soil, 14% cement and 80% residue. In terms of binary mixtures, we obtained the best result for the mix-cement residue, which was 14% cement incorporated in the residue

Estudo morfológico da blenda polimérica poli(metacrilato de metila)/poli(tereftalato de etileno) reciclado (PMMA/PET)

Among the options for plastics modification more convenient, both from a technical-scientific and economic, is the development of polymer blends by processing in the molten state. This work was divide into two stages, with the aim to study the phase morphology of binary blend PMMA / PET blend and this compatibilized by the addition of the poly(methyl methacrylate-co-glycidyl methacrylate-co-ethyl acrylate) copolymer (MMA-GMA-EA). In the first stage is analyzed the morphology of the blend at a preliminary stage where we used the bottle-grade PET in a Haake torque rheometer and the effect of compatibilizer in this blend was evaluated. In the second stage the blend was processed using the recycled PET in a single screw extruder and subsequently injection molding in the shape of specimens for mechanical tests. In both stages we used a transmission electron microscopy (TEM) to observe the morphologies of the samples and an image analyzer to characterize them. In the second stage, as well as analysis by TEM, tensile test, scanning electron microscopy (SEM) and atomic force microscopy (AFM) was performed to correlate the morphology with the mechanical properties. The samples used in morphological analyzes were sliced by cryo-ultramicrotomy technique for the analysis by TEM and the analysis by SEM and AFM, we used the flat face of the block after cut cryogenic. It was found that the size of the dispersed phase decreased with the addition of MMA-GMA-EA in blends prepared in a Haake. In the tensile test, the values of maximum tensile strength and modulus of elasticity is maintained in a range between the value of pure PMMA the pure PET, while the elongation at break was influenced by the composition by weight of the PMMA mixture. The coupling agent corroborated the results presented in the blend PMMA / PETrec / MMA-GMA-EA (80/15/5 %w/w), obtained by TEM, AFM and SEM. It was concluded that the techniques used had a good morphologic correlation, and can be confirmed for final analysis of the morphological characteristics of the blends PMMA / PET

Efeito da adição do poli(etileno acrilato de metila) (ema) e da fibra de linter de algodão nas propriedades do poli(tereftalato de etileno) reciclado (petrec)

The development of new materials to fill the demand of technological advances is a challenge for many researchers around the world. Strategies such as making blends and composites are promising alternatives to produce materials with different properties from those found in conventional polymers. The objective of this study is to evaluate the effect of adding the copolymer poly(ethylene methyl acrylate) (EMA) and cotton linter fibers (LB) on the properties of recycled poly(ethylene terephthalate) (PETrec) by the development of PETrec/EMA blend and PETrec/EMA/LB blend composite. In order to improve the properties of these materials were added as compatibilizers: Ethylene - methyl acrylate - glycidyl methacrylate terpolymer (EMA-GMA) and maleic anhydride grafted polyethylene (PE-g-MA). The samples were produced using a single screw extruder and then injection molded. The obtained materials were characterized by thermogravimetry (TG), melt flow index (MFI) mensurements, torque rheometry, pycnometry to determinate the density, tensile testing and scanning electron microscopy (SEM). The rheological results showed that the addition of the EMA copolymer increased the viscosity of the blend and LB reduces the viscosity of the blend composite. SEM analysis of the binary blend showed poor interfacial adhesion between the PETrec matrix and the EMA dispersed phase, as well as the blend composite of PETrec/EMA/LB also observed low adhesion with the LB fiber. The tensile tests showed that the increase of EMA percentage decreased the tensile strength and the Young s modulus, also lower EMA percentage samples had increased the elongation at break. The blend composite showed an increase in the tensile strength and in the Young`s modulus, and a decrease in the elongation at break. The blend formulations with lower EMA percentages showed better mechanical properties that agree with the particle size analysis which showed that these formulations presented a smaller diameter of the dispersed phase. The blend composite mechanical tests showed that this material is stronger and stiffer than the blend PETrec/EMA, whose properties have been reduced due to the presence of EMA rubbery phase. The use of EMA-GMA was effective in reducing the particle size of the EMA dispersed phase in the PETrec/EMA blend and PE-g-MA showed evidences of reaction with LB and physical mixture with the EMA

Testing the link specification in binary choice models. A semiparametric approach.

Doutoramento em Matemática.

Validation of neurobehavioral studies for evaluating the perinatal effects of single and mixture exposure to pesticides.

2009

The Effects of Instrumental Touching on Infant Pain Perception and the Effects of Eutectic Mixture of Local Anesthetics (EMLA) on the Reduction of Pain

Background: Premature infants, who have to spend the first week of their lives in neonatal intensive care units (NICUs), experience pain and stress in numerous cases, and they are exposed to many invasive interventions. The studies have shown that uncontrolled pain experienced during early life has negative and long-term side effects, such as distress, and such experiences negatively affect the development of the central nervous system Objectives: The purpose of the study was to examine the effects of touching on infant pain perception and the effects of eutectic mixture of local anesthetic (EMLA) on the reduction of pain. Patients and Methods: Data for the study were collected between March and August 2012 from the neonatal clinic of a university hospital located in eastern Turkey. The population of the study consisted of premature infants who were undergoing treatment, completed the first month and who were approved for Hepatitis B vaccine. The study consisted of two experimental groups and one control group. Information forms, intervention follow-up forms, and Premature Infant Pain Profile (PIPP) were used to collect the data. EMLA cream was applied on the vastus lateralis muscles of the first experimental group before the vaccination. The second experimental group was vaccinated by imitation (placebo), without a needle tip or medicine. Vaccination was carried out using instrumental touch in this group. A routine vaccination was applied in the control group. Results: Mean pain scores of the group to which EMLA was applied were lower in a statistically significant way (P < 0.05) compared to the pain scores of the other groups. Moreover, it was determined that even though invasive intervention was not applied to the newborns, the touching caused them to feel pain just as in the placebo group (P < 0.005). Conclusions: The results demonstrated that EMLA was an effective method for reducing pain in premature newborns, and the use of instrumental touch for invasive intervention stimulated the pain perception in the newborns.

What to do when K-means clustering fails:a simple yet principled alternative algorithm

The K-means algorithm is one of the most popular clustering algorithms in current use as it is relatively fast yet simple to understand and deploy in practice. Nevertheless, its use entails certain restrictive assumptions about the data, the negative consequences of which are not always immediately apparent, as we demonstrate. While more flexible algorithms have been developed, their widespread use has been hindered by their computational and technical complexity. Motivated by these considerations, we present a flexible alternative to K-means that relaxes most of the assumptions, whilst remaining almost as fast and simple. This novel algorithm which we call MAP-DP (maximum a-posteriori Dirichlet process mixtures), is statistically rigorous as it is based on nonparametric Bayesian Dirichlet process mixture modeling. This approach allows us to overcome most of the limitations imposed by K-means. The number of clusters K is estimated from the data instead of being fixed a-priori as in K-means. In addition, while K-means is restricted to continuous data, the MAP-DP framework can be applied to many kinds of data, for example, binary, count or ordinal data. Also, it can efficiently separate outliers from the data. This additional flexibility does not incur a significant computational overhead compared to K-means with MAP-DP convergence typically achieved in the order of seconds for many practical problems. Finally, in contrast to K-means, since the algorithm is based on an underlying statistical model, the MAP-DP framework can deal with missing data and enables model testing such as cross validation in a principled way. We demonstrate the simplicity and effectiveness of this algorithm on the health informatics problem of clinical sub-typing in a cluster of diseases known as parkinsonism.

Performance of concrete incorporating amorphous silica residue and biomass fly ash

L'industrie du ciment est l'une des principales sources d'émission de dioxyde de carbone. L'industrie mondiale du ciment contribue à environ 7% des émissions de gaz à effet de serre dans l'atmosphère. Afin d'aborder les effets environnementaux associés à la fabrication de ciment exploitant en permanence les ressources naturelles, il est nécessaire de développer des liants alternatifs pour fabriquer du béton durable. Ainsi, de nombreux sous-produits industriels ont été utilisés pour remplacer partiellement le ciment dans le béton afin de générer plus d'économie et de durabilité. La performance d'un additif de ciment est dans la cinétique d'hydratation et de la synergie entre les additions et de ciment Portland. Dans ce projet, deux sous-produits industriels sont étudiés comme des matériaux cimentaires alternatifs: le résidu de silice amorphe (RSA) et les cendres des boues de désencrage. Le RSA est un sous-produit de la production de magnésium provenant de l'Alliance Magnésium des villes d'Asbestos et Thedford Mines, et les cendres des boues de désencrage est un sous-produit de la combustion des boues de désencrage, l'écorce et les résidus de bois dans le système à lit fluidisé de l'usine de Brompton située près de Sherbrooke, Québec, Canada. Récemment, les cendres des boues de désencrage ont été utilisées comme des matériaux cimentaires alternatifs. L'utilisation de ces cendres comme matériau cimentaire dans la fabrication du béton conduit à réduire la qualité des bétons. Ces problèmes sont causés par des produits d'hydratation perturbateurs des cendres volantes de la biomasse quand ces cendres sont partiellement mélangées avec du ciment dans la fabrication du béton. Le processus de pré-mouillage de la cendre de boue de désencrage avant la fabrication du béton réduit les produits d'hydratation perturbateurs et par conséquent les propriétés mécaniques du béton sont améliorées. Les approches pour étudier la cendre de boue de désencrage dans ce projet sont : 1) caractérisation de cette cendre volante régulière et pré-humidifiée, 2) l'étude de la performance du mortier et du béton incorporant cette cendre volante régulière et pré-humidifiée. Le RSA est un nouveau sous-produit industriel. La haute teneur en silice amorphe en RSA est un excellent potentiel en tant que matériau cimentaire dans le béton. Dans ce projet, l'évaluation des RSA comme matériaux cimentaires alternatifs compose trois étapes. Tout d'abord, la caractérisation par la détermination des propriétés minéralogiques, physiques et chimiques des RSA, ensuite, l'optimisation du taux de remplacement du ciment par le RSA dans le mortier, et enfin l'évaluation du RSA en remplacement partiel du ciment dans différents types de béton dans le système binaire et ternaire. Cette étude a révélé que le béton de haute performance (BHP) incorporant le RSA a montré des propriétés mécaniques et la durabilité, similaire du contrôle. Le RSA a amélioré les propriétés des mécaniques et la durabilité du béton ordinaire (BO). Le béton autoplaçant (BAP) incorporant le RSA est stable, homogène et a montré de bonnes propriétés mécaniques et la durabilité. Le RSA avait une bonne synergie en combinaison de liant ternaire avec d'autres matériaux cimentaires supplémentaires. Cette étude a montré que le RSA peut être utilisé comme nouveaux matériaux cimentaires dans le béton.

The dehydrogenation reactions between LiH and organic amines for solid state hydrogen storage systems

Hydrogen is considered as an appealing alternative to fossil fuels in the pursuit of sustainable, secure and prosperous growth in the UK and abroad. However there exists a persisting bottleneck in the effective storage of hydrogen for mobile applications in order to facilitate a wide implementation of hydrogen fuel cells in the fossil fuel dependent transportation industry. To address this issue, new means of solid state chemical hydrogen storage are proposed in this thesis. This involves the coupling of LiH with three different organic amines: melamine, urea and dicyandiamide. In principle, thermodynamically favourable hydrogen release from these systems proceeds via the deprotonation of the protic N-H moieties by the hydridic metal hydride. Simultaneously hydrogen kinetics is expected to be enhanced over heavier hydrides by incorporating lithium ions in the proposed binary hydrogen storage systems. Whilst the concept has been successfully demonstrated by the results obtained in this work, it was observed that optimising the ball milling conditions is central in promoting hydrogen desorption in the proposed systems. The theoretical amount of 6.97 wt% by dry mass of hydrogen was released when heating a ball milled mixture of LiH and melamine (6:1 stoichiometry) to 320 °C. It was observed that ball milling introduces a disruption in the intermolecular hydrogen bonding network that exists in pristine melamine. This effect extends to a molecular level electron redistribution observed as a function of shifting IR bands. It was postulated that stable phases form during the first stages of dehydrogenation which contain the triazine skeleton. Dehydrogenation of this system yields a solid product Li2NCN, which has been rehydrogenated back to melamine via hydrolysis under weak acidic conditions. On the other hand, the LiH and urea system (4:1 stoichiometry) desorbed approximately 5.8 wt% of hydrogen, from the theoretical capacity of 8.78 wt% (dry mass), by 270 °C accompanied by undesirable ammonia and trace amount of water release. The thermal dehydrogenation proceeds via the formation of Li(HN(CO)NH2) at 104.5 °C; which then decomposes to LiOCN and unidentified phases containing C-N moieties by 230 °C. The final products are Li2NCN and Li2O (270 °C) with LiCN and Li2CO3 also detected under certain conditions. It was observed that ball milling can effectively supress ammonia formation. Furthermore results obtained from energetic ball milling experiments have indicated that the barrier to full dehydrogenation between LiH and urea is principally kinetic. Finally the dehydrogenation reaction between LiH and dicyandiamide system (4:1 stoichiometry) occurs through two distinct pathways dependent on the ball milling conditions. When ball milled at 450 RPM for 1 h, dehydrogenation proceeds alongside dicyandiamide condensation by 400 °C whilst at a slower milling speed of 400 RPM for 6h, decomposition occurs via a rapid gas desorption (H2 and NH3) at 85 °C accompanied by sample foaming. The reactant dicyandiamide can be generated by hydrolysis using the product Li2NCN.