Predictive model for growth of Salmonella Newport on Romaine lettuce

Cross-contamination of ready-to-eat (RTE) salad vegetables with Salmonella from raw chicken followed by growth during meal preparation are important risk factors for human salmonellosis. To better predict and manage this risk, a model (general regression neural network) for growth of a chicken isolate of Salmonella Newport (0.91 log) on Romaine lettuce (0.18 g) at times (0–8 hr) and temperatures (16–40°C) observed during meal preparation was developed with Excel, NeuralTools, and @Risk. Model performance was evaluated using the acceptable prediction zones (APZ) method. The proportion of residuals in the APZ (pAPZ) was 0.93 for dependent data (n = 210) and 0.93 for independent data (n = 72) for interpolation. A pAPZ ≥0.70 indicates acceptable model performance. Thus, the model was successfully validated for interpolation and can be used with confidence to predict and manage this important risk to public health.     

Triblock SEBS/DVB crosslinked and sulfonated membranes: Fuel cell performance and conductivity

A set of styrene-ethylene-butylene-styrene triblock copolymer (SEBS) membranes with 10 or 25 wt% divinyl-benzene (DVB) as a crosslinking agent were prepared and validated. Physicochemical characterization revealed suitable hydrolytic and thermal stability of photo-crosslinked membranes containing 25 wt% DVB and post-sulfonated. These compositions were evaluated in H₂/O₂ single cells, and electrical and proton conductivities were furtherly assessed. The membranes with the milder post-sulfonation showed greater proton conductivity than those with excessive sulfonation. In terms of electrical conductivity, a universal power law was applied, and the values obtained were low enough for being used as polyelectrolytes. At the analyzed temperatures, the charge transport process follows a long-range pathway or vehicular model. Finally, fuel cell performance revealed the best behavior for the membrane with 25 wt% DVB, photo-crosslinked during 30 min and mild sulfonated, with a promising power density of 526 mW·cm⁻². Overall, the results obtained highlight the promising fuel cell performance of these cost-effective triblock copolymer-based membranes and indicate that higher sulfonation does not necessarily imply better power density.     

DSVD‐autoencoder: A scalable distributed privacy‐preserving method for one‐class classification

One‐class classification has gained interest as a solution to certain kinds of problems typical in a wide variety of real environments like anomaly or novelty detection. Autoencoder is the type of neural network that has been widely applied in these one‐class problems. In the Big Data era, new challenges have arisen, mainly related with the data volume. Another main concern derives from Privacy issues when data is distributed and cannot be shared among locations. These two conditions make many of the classic and brilliant methods not applicable. In this paper, we present distributed singular value decomposition (DSVD‐autoencoder), a method for autoencoders that allows learning in distributed scenarios without sharing raw data. Additionally, to guarantee privacy, it is noniterative and hyperparameter‐free, two interesting characteristics when dealing with Big Data. In comparison with the state of the art, results demonstrate that DSVD‐autoencoder provides a highly competitive solution to deal with very large data sets by reducing training from several hours to seconds while maintaining good accuracy.     

Enabling fast and energy-efficient FM-index exact matching using processing-near-memory

The Journal of Supercomputing - Memory bandwidth and latency constitutes a major performance bottleneck for many data-intensive applications. While high-locality access patterns take advantage of...     

Effect of 1-D and 2-D carbon-based nano-reinforcements on the dry sliding-wear behaviour of 3Y-TZP ceramics

The dry sliding-wear behaviour of a zirconia doped with 3 mol% yttria (3Y-TZP) monolithic and of two 3Y-TZP composites, these latter with the same ultrafine-grained microstructure as the former but reinforced with either 1-D carbon nanofibres (CNFs) or 2-D reduced graphene oxide (rGO) nanoplatelets, was evaluated by ball-on-disk tests at moderate load, and compared critically. It was found that 3Y-TZP, 3Y-TZP/CNFs, and 3Y-TZP/rGO undergo mild wear, in the three cases by abrasion with contributions from both plastic deformation and fracture (with varying severities depending on the sample). It was also found that wear resistance does not correlate with hardness or toughness, but with whether or not there is formation of self-lubricating tribofilms on the contact surface. Specifically, once pulled-out, 2-D rGO nano-reinforcements impose solid-state lubrication that reduces the coefficient of friction (CoF), thus providing 3Y-TZP/rGO with superior wear resistance relative to both 3Y-TZP and 3Y-TZP/CNFs. 1-D CNF nano-reinforcements, however, do not form such tribofilms, or hardly do so, thus having no effect on the CoF and wear resistance. Implications are discussed of both the dimensionality of the carbon-based nano-reinforcements and the testing conditions for the microstructural design of ceramic composites for tribological applications.     

A duality relationship between fuzzy metrics and metrics

Based on the duality relationship between indistinguishability operators and (pseudo-)metrics, we address the problem of establishing whether there is a relationship between the last ones and fuzzy (pseudo-)metrics. We give a positive answer to the posed question. Concretely, we yield a method for generating fuzzy (pseudo-)metrics from (pseudo)-metrics and vice versa. The aforementioned methods involve the use of the pseudo-inverse of the additive generator of a continuous Archimedean t-norm. As a consequence, we get a method to generate non-strong fuzzy (pseudo-)metrics from (pseudo-)metrics. Examples that illustrate the exposed methods are also given. Finally, we show that the classical duality relationship between indistinguishability operators and (pseudo)-metrics can be retrieved as a particular case of our results when continuous Archimedean t-norms are under consideration.     

Metallurgical effects and distortions in laser welding of thin sheet steels with variations in strength

Geometrical distortions occur while welding, but the understanding of how and why they occur and how to control them is limited. The relation between the weld width, weld metal volume, total energy input, width of hard zone and distortions when laser welding three different thin sheet steels with varying strength has therefore been studied. Weld metal volume and total energy input show a good correlation with distortion for one steel at a time. The best correlation with the when including all three steel grades was the width of the hard zone composed of weld metal and the martensitic area in the heat affected zone.     

Toward a software transactional memory for heterogeneous CPU–GPU processors

The Journal of Supercomputing - The heterogeneous accelerated processing units (APUs) integrate a multi-core CPU and a GPU within the same chip. Modern APUs implement CPU–GPU platform atomics...