Optimal inspection for randomly triggered hidden deterioration processes

This paper deals with degradation processes whose onset is triggered at a random time and which stay hidden until they are discovered through inspection or when they begin to show symptoms. This is applicable in many healthcare and industrial scenarios, for example, in the modeling of breast cancer or termite infestation. In our model, we assume that symptoms appear after hitting a random critical threshold and that inspections may have a sensitivity less than one as well as a nonzero false positive rate. The expected cost of repair is derived, and the inspection rate is optimized for a cycle (which lasts from degradation-free to repaired state). This gives results for three cases: the first is for a finite observation period with no degradation recurrence, the second for infinite time horizon allowing recurrence. In the third case, we derive an upper bound for the expected cost in a given constant time period. Finally, the model is applied to determine the optimal strategy for breast cancer screening with regard to the effects of different parametrizations.     

R2D2: A scalable deep learning toolkit for medical imaging segmentation

Deep learning has gained a significant popularity in recent years thanks to its tremendous success across a wide range of relevant fields of applications, including medical image analysis domain in particular. Although convolutional neural networks (CNNs) based medical applications have been providing powerful solutions and revolutionizing medicine, efficiently training of CNNs models is a tedious and challenging task. It is a computationally intensive process taking long time and rare system resources, which represents a significant hindrance to scientific research progress. In order to address this challenge, we propose in this article, R2D2, a scalable intuitive deep learning toolkit for medical imaging semantic segmentation. To the best of our knowledge, the present work is the first that aims to tackle this issue by offering a novel distributed versions of two well-known and widely used CNN segmentation architectures [ie, fully convolutional network (FCN) and U-Net]. We introduce the design and the core building blocks of R2D2. We further present and analyze its experimental evaluation results on two different concrete medical imaging segmentation use cases. R2D2 achieves up to 17.5× and 10.4× speedup than single-node based training of U-Net and FCN, respectively, with a negligible, though still unexpected segmentation accuracy loss. R2D2 offers not only an empirical evidence and investigates in-depth the latest published works but also it facilitates and significantly reduces the effort required by researchers to quickly prototype and easily discover cutting-edge CNN configurations and architectures.     

The Electron–Phonon Interaction of Low-Dimensional and Multi-Dimensional Materials from He Atom Scattering

Atom scattering is becoming recognized as a sensitive probe of the electron–phonon interaction parameter λ at metal and metal-overlayer surfaces. Here, the theory is developed, linking λ to the thermal attenuation of atom scattering spectra (in particular, the Debye–Waller factor), to conducting materials of different dimensions, from quasi-1D systems such as W(110):H(1 × 1) and Bi(114), to quasi-2D layered chalcogenides, and high-dimensional surfaces such as quasicrystalline 2ML-Ba(0001)/Cu(001) and d-AlNiCo(00001). Values of λ obtained using He atoms compare favorably with known values for the bulk materials. The corresponding analysis indicates in addition, the number of layers contributing to the electron–phonon interaction, which is measured in an atom surface collision.     

The Brazilian academic genealogy: evidence of advisor–advisee relationships through quantitative analysis

Scientometrics - Science can be examined from several standpoints, such as through a bibliometric analysis of the scientific output of researchers, research groups or institutions. However, there...     

Biomimetic Omnidirectional Antireflective Glass via Direct Ultrafast Laser Nanostructuring

Here, a single‐step, biomimetic approach for the realization of omnidirectional transparent antireflective glass is reported. In particular, it is shown that circularly polarized ultrashort laser pulses produce self‐organized nanopillar structures on fused silica (SiO₂). The laser‐induced nanostructures are selectively textured on the glass surface in order to mimic the spatial randomness, pillar‐like morphology, as well as the remarkable antireflection properties found on the wings of the glasswing butterfly, Greta oto, and various Cicada species. The artificial structures exhibit impressive antireflective properties, both in the visible and infrared frequency ranges, which are remarkably stable over time. Accordingly, the laser‐processed glass surfaces show reflectivity smaller than 1% for various angles of incidence in the visible spectrum for s–p linearly polarized configurations. However, in the near‐infrared spectrum, the laser‐textured glass shows higher transmittance compared to the pristine. It is envisaged that the current results will revolutionize the technology of antireflective transparent surfaces and impact numerous applications from glass displays to optoelectronic devices.     

ARTMe: a new array-based algorithm for Adaptive Refinement of Triangle Meshes

Engineering with Computers - This work presents a new efficient array-based algorithm for adaptive mesh refinement capable of interactively generating millions of triangles. The new refinement...