Dynamic recurrent neural networks: a dynamical analysis

In this paper, we explore the dynamical features of a neural network model which presents two types of adaptative parameters: the classical weights between the units and the time constants associated with each artificial neuron. The purpose of this study is to provide a strong theoretical basis for modeling and simulating dynamic recurrent neural networks. In order to achieve this, we study the effect of the statistical distribution of the weights and of the time constants on the network dynamics and we make a statistical analysis of the neural transformation. We examine the network power spectra (to draw some conclusions over the frequential behaviour of the network) and we compute the stability regions to explore the stability of the model. We show that the network is sensitive to the variations of the mean values of the weights and the time constants (because of the temporal aspects of the learned tasks). Nevertheless, our results highlight the improvements in the network dynamics due to the introduction of adaptative time constants and indicate that dynamic recurrent neural networks can bring new powerful features in the field of neural computing.     

Selective removal of dissolved uranium in drinking water by nanofiltration

A procedure for the selective removal of uranium traces dissolved in drinking water has been studied. Plate module membrane filtration equipment was operated to evaluate the performance and selectivity of three different nanofiltration flat-sheet membranes. Experiments were carried out using various commercial mineral waters with distinct physicochemical compositions. The membranes were first discriminating by their ability to reject uranium in the presence of the main cations found in mineral waters, using a 2 mg L(-1) (2000 ppb) concentration of uranium. The rejection of U(VI) was dependent on the uranyl speciation and the ionic strength. Second, removal of uranium traces (0.02 mg L(-1), 20 ppb) was performed using the nanofiltration membrane showing the highest selectivity for uranium toward alkaline and alkaline-earth ions. The results showed a high performance of the nanofiltration membrane, Osmonics DL, for selective uranium rejection at low pressure (1 bar), illustrating the advantage of nanofiltration for the selective removal of uranium from drinking water.     

Influence of hydrogenation and antioxidants on the stability of soybean oil biodiesels

This study examined the influence of hydrogenation and antioxidant activity of natural products derived from Mangifera indica on soybean biodiesel stability. Biodiesels were prepared with refined, semi‐refined, and partially hydrogenated soybean oil and the stability was evaluated in a Rancimat apparatus. The biodiesel from partially hydrogenated soybean oil showed greater stability (11.9 h), exceeding the standard limit established by the Brazilian National Petroleum Agency (6 h). Gallic acid, tannic acid, and mango seed kernel extract were evaluated as antioxidants. The gallic acid showed higher antioxidant ability than the other products, increasing twofold the oxidative stability (20.8 h). The results indicate the importance of mango seed as a source of efficient antioxidants for biodiesel just like gallic acid. Practical applications: The results provide detailed information about: (i) the influence of hydrogenation on soybean biodiesel stability; (ii) the antioxidant activity of natural products derived from M. indica; (iii) and the importance of each phenolic constituent on the stability. Mango seed is an abundant waste product from fruit juice industry in Brazil and could be used for obtaining an effective antioxidant additive for biodiesel.     

Safety assessment of dairy microorganisms: the Lactobacillus genus

Lactobacilli are Gram positive rods belonging to the Lactic Acid Bacteria (LAB) group. Their phenotypic traits, such as each species' obligate/facultative, homo/heterofermentation abilities play a crucial role in souring raw milk and in the production of fermented dairy products such as cheese, yoghurt and fermented milk (including probiotics). An up to date safety analysis of these lactobacilli is needed to ensure consumer safety. Lactobacillus genus is a heterogeneous microbial group containing some 135 species and 27 subspecies, whose classification is constantly being reshuffled. With the recent use of advanced molecular methods it has been suggested that the extreme diversity of the Lactobacillus genomes would justify recognition of new subgeneric divisions. A combination of genotypic and phenotypic tests, for example DNA-based techniques and conventional carbohydrate tests, is required to determine species. Pulsed-Field gel Electrophoresis (PFGE) has been successfully applied to strains of dairy origin and is the most discriminatory and reproducible method for differentiating Lactobacillus strains. The bibliographical data support the hypothesis that the ingestion of Lactobacillus is not at all hazardous since lactobacillemia induced by food, particularly fermented dairy products, is extremely rare and only occurs in predisposed patients. Some metabolic features such as the possible production of biogenic amines in fermented products could generate undesirable adverse effects. A minority of starter and adjunct cultures and probiotic Lactobacillus strains may exceptionally show transferable antibiotic resistance. However, this may be underestimated as transferability studies are not systematic. We consider that transferable antibiotic resistance is the only relevant cause for caution and justifies performing antibiotic-susceptibility assays as these strains have the potential to serve as hosts of antibiotic-resistance genes, with the risk of transferring these genes to other bacteria. However, as a general rule, lactobacilli have a high natural resistance to many antibiotics, especially vancomycin, that is not transferable. Safety assessment requirements for Lactobacillus strains of technological interest should be limited to an antibiotic profile and a study to determine whether any antibiotic resistance(s) of medical interest detected is (or are) transferable. This agrees with the recent EFSA proposal suggesting attribution of a QPS status for 32 selected species of lactobacilli.     

Etude de verres de chalcogenures contenant de l'europium divalent systeme EuSGa2S3GeS2

Existence of a very large glassy domain inside the ternary system EuSGa₂S₃GeS₂. Conditions of formation of the glassy products. Thermal properties : glass trarsition temperatures and crystallization temperatures. Existence in the ternary system of a large region of glass having a very small crystallization speed. Refractiv index. Magnetic moment. Structural study by Raman spectroscopy of crystalline compounds (GeS₂, EuGeS₃, Eu₂GeS₄, EuGa₂S₄) and of various glasses of this system.     

Adaptative time constants improve the prediction capability of recurrent neural networks

Classical statistical techniques for prediction reach their limitations in applications with nonlinearities in the data set; nevertheless, neural models can counteract these limitations. In this paper, we present a recurrent neural model where we associate an adaptative time constant to each neuron-like unit and a learning algorithm to train these dynamic recurrent networks. We test the network by training it to predict the Mackey-Glass chaotic signal. To evaluate the quality of the prediction, we computed the power spectra of the two signals and computed the associated fractional error. Results show that the introduction of adaptative time constants associated to each neuron of a recurrent network improves the quality of the prediction and the dynamical features of a neural model. The performance of such dynamic recurrent neural networks outperform time-delay neural networks.     

Optimization of Extreme Load and Break-in Period in Plain ZDDP Oil with FeF3 Catalyst Using Design of Experiment and Fundamental Study under Different Speeds

The mechanism of tribofilm formation and breakdown was carefully followed and studied in 0.1 P% (percentage phosphorus content) plain zinc dialkyldithiophosphate (ZDDP) oil in the presence of iron fluoride (FeF₃) catalyst under extreme Hertzian contact pressure (3.0 GPa) and two different rotational speeds or variable speed with break in period (100 rpm for the first 5,000 revolutions and a 700 rpm until failure or 100,000 revolutions, whichever comes first). At the onset of large frictional fluctuations, the contact surface temperature increased significantly and reached 90°C ± 5°C. The present article describes an innovative method of reducing the surface temperature by using a break in period of 2 or 3 min and rerunning the test until failure. The two different rotational speeds or variable speed will be compared to a steady-state speed of 700 rpm.

Thermal decomposition of ZDDP is examined in the presence of powder and dispersed FeF₃ using differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA). Surface analyses were carried out using scanning electron microscopy (SEM) coupled with X-ray of the wear tracks, transmission electron microscopy (TEM), and auger electron spectroscopy (AES). Results showed that submicrometer dispersed FeF₃ provided excellent wear protection when combined with ZDDP in the variable-speed test with break-in by forming a tribofilm that is amorphous in nature and rich in phosphorus, which was shown by the TEM and X-ray analyses.     

Cloud point and solvent extraction study of uranium(VI) by 8‐hydroxyquinoline

BACKGROUND: Cloud point extraction (CPE) is an attractive alternative to solvent extraction. However, comparisons between both techniques are lacking. In this paper, the extraction of uranium(VI) using 8‐hydroxyquinoline (HQ) as chelating agent was studied by CPE using Triton X‐114 as non‐ionic surfactant and by solvent extraction using CHCl₃ as diluent. RESULTS: Using CPE, a quantitative extraction was observed for pH higher than 4.5 with a HQ/U ratio of 10. Using solvent extraction an increase in the HQ/U ratio up to 50 is necessary to obtain a near quantitative extraction. Both extraction systems were then compared with respect to the nature of extracted species, and the extraction constants determined using log‐log analysis of the extraction data. In the solvent extraction system, the extracted species were identified as UO₂Q₂ and the corresponding extraction constant was found to be log k_ex = ? 3.6 ± 0.2 on the molar scale. Considering that UO₂Q₂ is also the extracted species in CPE, a slightly higher extraction constant, i.e. log k_ex = ? 2.5 ± 0.3, was found. CONCLUSION: Such a small difference in favour of the CPE system may arise from the combination of various phenomena, including effects of temperature and effects of ‘extractant environment’. However, a change in the nature of the extracted species, namely from UO₂Q₂ in the solvent extraction system to the formation of adducts, i.e. UO₂Q₂(HQ) and UO₂Q₂(HQ)₂ in the CPE system, due to higher HQ concentration in the surfactant‐rich phase compared with chloroform, cannot be precluded, but requires confirmation. Copyright © 2012 Society of Chemical Industry     

Sequence of a 17·1 kb DNA fragment from chromosome X of Saccharomyces cerevisiae includes the mitochondrial ribosomal protein L8

We have sequenced a continuous segment of 17 137 bp on chromosome X. Sequence analysis of this stretch revealed 14 open reading frames (ORFs) at least 100 amino acids long. One gene, encoding the mitochondrial 60S ribosomal protein L8, had already been sequenced. Four ORF products show weak homologies with known protein sequences. The nine remaining ORF products have no homologies with sequences in data banks. The nucleotide sequence of the 17·1 kb fragment is available through the EMBL data library under Accession Number Z34288.