Search

In this article, we consider the problem of privacy-preserving distributed deep learning where data privacy is protected by fully homomorphic encryption. The aim is to develop a method for practical and scalable distributed deep learning with fully homomorphic encrypted data. The method must address the issue arising from the large computational cost associated with fully homomorphic encrypted data to offer a practical and scalable solution. An approach that leverages fuzzy-based membership mappings for data representation learning is considered for distributed deep learning with fully homomorphic encrypted data. The method introduces globally convergent and robust variational membership mappings to build local deep models. The local models are combined in a robust and flexible manner by means of fuzzy attributes to build a global model such that the global model can be homomorphically evaluated in an efficient manner. The membership-mappings-based privacy-preserving distributed deep learning method is accurate, practical, and scalable. This is verified through numerous experiments that include demonstrations using MNIST and Freiburg Groceries datasets, and a biomedical application related to the detection of mental stress on individuals. This study develops globally convergent and robust variational membership mappings for their application to accurate, practical, and scalable privacy-preserving distributed deep learning.

Artificial or human test bones are used for the biomechanical testing of implants. Human test bones are rare and not always available. These must, therefore, be substituted with artificial test bones. However, current artificial test bones are only available with specific characteristics (e.g., age groups or disease characteristics). Additionally, their mechanical properties are only comparable to a limited extent to those of a human bone. This paper presents a methodology for designing additively manufactured artificial test bones for biomechanical testing that replicate the mechanical behavior of a human bone. Topology optimization methods are used to generate the artificial test bone's internal structure. The geometric model is based on a computed tomography dataset of a human bone. The input data can be manipulated in advance to reproduce defects or disease patterns. The bone was fixed at the distal diaphysis and loaded with different biomechanical forces for topology optimization. Boundary conditions due to possible additive manufacturing processes were incorporated into the optimization to ensure manufacturability. The optimization result is compared with experimental data from a human bone. A bone-like internal structure and increased compliance of the topology-optimized test bone model compared to the commercial model were observed.

This paper presents the results of a case study regarding understanding frugal engineering design and process methodology through the fabrication of wearable tent-jackets for the protection of homeless individuals. Frugal innovations are being investigated for socially conscious products, but an understanding of frugal engineering methodology remains unexplored and is the focus of this research. Multiple socio-economic drivers are resulting in a lack of access to basic human needs such as housing, healthcare, etc. across the globe. Frugal engineering presents an equitable and practical engineering approach to address these challenges and merits exploring a methodology for innovations. In this paper, an origami-based design approach was employed to create a functional prototype of a tent for shelter that can be morphed into a wearable weather protection jacket. We call it a “wearable tent”. The prototype was successfully fabricated using a reusable material that is used abundantly in the construction industry. The functional prototype was tested for its efficacy and effectiveness in terms of weight, strength, and ease of switching between the two configurations. The results suggest a potential model that could be explored and applied for socially conscious engineering design solutions for such severely resource-constrained cases for disadvantaged communities with future potential for digitally disadvantaged enterprises.

Electron correlations amplify quantum fluctuations and, as such, are recognized as the origin of many quantum phases. However, whether strong correlations can lead to gapless topological states is an outstanding question, in part because many of the ideas in topological condensed-matter physics rely on the analysis of an effectively non-interacting band structure. Therefore, a framework that allows the identification of strongly correlated topological materials is needed. Here we suggest a general approach in which strong correlations cooperate with crystalline symmetry to drive gapless topological states. We test this materials design principle by exploring Kondo lattice models and materials whose space-group symmetries promote different kinds of electronic degeneracies. This approach allows us to identify Weyl–Kondo nodal-line semimetals with nodes pinned to the Fermi energy, demonstrating that it can be applied to discover strongly correlated topological semimetals. We identify three heavy-fermion compounds as material candidates, provide direct experimental evidence for our prediction in Ce2Au3In5and discuss how our approach may lead to many more. Our findings illustrate the potential of this materials design principle to guide the search for new topological metals in a broad range of strongly correlated systems.

Here we characterize the new Innsbruck wall free impinging jets reactor (INNpinJeR) and compare its performance with the TROPOS free jet flow system by quantifying oxidation products of the well-understood ozonolysis reactions of tetramethyl ethylene (TME) and cyclohexene with an ammonium chemical ionization mass spectrometer. We support the results with flow and chemical simulations on the k- RANS steady-state flow field. Flow simulations show a mixing region in the center of the reactor that does not disturb the surrounding laminar main flow. Quantitative detection of several peroxy radicals strongly suggests negligible wall contacts. Direct comparison with the TROPOS flow system using cyclohexene ozonolysis with the same analytical technique shows very good reproducibility. The total effective reaction time amounts to 9.0–9.3 seconds found from the flow and chemistry simulations and 9.4 ± 1.1 seconds according to the analysis of TME ozonolysis products. These features allow for characterizing the sensitivity of analytical instruments towards different oxidation products, including RO2radicals, but most importantly we can conclude, that the Innsbruck wall-free INNpinJeR can be used to study gas-phase chemistry under atmospherically relevant conditions without significant influence of interface chemistry.

Nanoscopic magnetic domain walls (DWs), via their absence or presence, enable highly interesting binary data bits. The current-controlled, high-speed, synchronous motion of sequences of chiral DWs in magnetic nanoconduits induced by current pulses makes possible high-performance spintronic memory and logic devices. The closer the spacing between neighboring DWs in an individual conduit or nanowire, the higher the data density of the device, but at the same time, the more difficult it is to read the bits. Here, we show how the DW spacing can be dynamically varied to facilitate reading for otherwise closely packed bits. In the first method, the current density is increased in portions of the conduit that, thereby, locally speeds up the DWs, decompressing them and making them easier to read. In the second method, a localized bias current is used to compress and decompress the DW spacing. Both of these methods are demonstrated experimentally and validated by micromagnetic simulations. DW compression and decompression rates as high as 88% are shown. These methods can increase the density with which DWs can be packed in future DW-based spintronic devices by more than an order of magnitude.

We developed a novel method to fabricate copper nanorods in situin a poly(ether sulfone) (15 wt %) casting solution by a sonochemical reduction of Cu2+ions with NaBH4. The main twist is the addition of ethanol to remove excess NaBH4through Cu(0) catalyzed ethanolysis. This enabled the direct use of the resulting copper-containing casting dispersions for membrane preparation by liquid nonsolvent-induced phase separation and led to full utilization of the copper source, generating zero metal waste. We characterized the copper nanorods as presented in the membranes via scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDX), transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD), and UV/vis spectroscopy. We could demonstrate that the rapid immobilization from reducing conditions led to the membrane incorporation of copper nanorods in a state of high reactivity, which also promoted the complete oxidation to CuO after fabrication. We further observed a large aspect ratio and crystal straining of the nanorods, likely resulting from growth around the matrix polymer. The entanglement with poly(ether sulfone) further facilitated a selective presentation at the pore surface of the final CuO-decorated membranes. The membranes also exhibit high water permeances of up to 2800 L/m2hbar. Our catalytic membranes achieved exceptionally high activities in the aqueous flow-through reduction of p-nitrophenol (p-NP), with turnover frequencies of up to 115 h–1, even surpassing those of other state-of-the-art catalytic membranes that incorporate Pd or Ag. Additionally, we demonstrated that catalytic hydrolysis of the reducing agent in water can lead to hydrogen gas formation and blocking of active sites during continuous catalytic p-NP hydrogenation. We illustrated that the accompanying conversion loss can be mitigated by facilitated gas transport in the water-filled pores, which is dependent on the orientation of the pore size gradient and the flow direction.

To better understand the role of aromatic hydrocarbons in new-particle formation, we measured the particle-phase abundance and volatility of oxidation products following the reaction of aromatic hydrocarbons with OH radicals. For this we used thermal desorption in an iodide-adduct Time-of-Flight Chemical-Ionization Mass Spectrometer equipped with a Filter Inlet for Gases and AEROsols (FIGAERO-ToF-CIMS). The particle-phase volatility measurements confirm that oxidation products of toluene and naphthalene can contribute to the initial growth of newly formed particles. Toluene-derived (C7) oxidation products have a similar volatility distribution to that of α-pinene-derived (C10) oxidation products, while naphthalene-derived (C10) oxidation products are much less volatile than those from toluene or α-pinene; they are thus stronger contributors to growth. Rapid progression through multiple generations of oxidation is more pronounced in toluene and naphthalene than in α-pinene, resulting in more oxidation but also favoring functional groups with much lower volatility per added oxygen atom, such as hydroxyl and carboxylic groups instead of hydroperoxide groups. Under conditions typical of polluted urban settings, naphthalene may well contribute to nucleation and the growth of the smallest particles, whereas the more abundant alkyl benzenes may overtake naphthalene once the particles have grown beyond the point where the Kelvin effect strongly influences the condensation driving force.

We use a real-time temperature-programmed desorption chemical-ionization mass spectrometer (FIGAERO–CIMS) to measure particle-phase composition and volatility of nucleated particles, studying pure α-pinene oxidation over a wide temperature range (−50 °C to +25 °C) in the CLOUD chamber at CERN. Highly oxygenated organic molecules are much more abundant in particles formed at higher temperatures, shifting the compounds toward higher O/C and lower intrinsic (300 K) volatility. We find that pure biogenic nucleation and growth depends only weakly on temperature. This is because the positive temperature dependence of degree of oxidation (and polarity) and the negative temperature dependence of volatility counteract each other. Unlike prior work that relied on estimated volatility, we directly measure volatility via calibrated temperature-programmed desorption. Our particle-phase measurements are consistent with gas-phase results and indicate that during new-particle formation from α-pinene oxidation, gas-phase chemistry directly determines the properties of materials in the condensed phase. We now have consistency between measured gas-phase product concentrations, product volatility, measured and modeled growth rates, and the particle composition over most temperatures found in the troposphere.

α-Pinene (C10H16) represents one of the most important biogenic emissions in the atmosphere. Its oxidation products can significantly contribute to the secondary organic aerosol (SOA) formation. Here, we report on the formation mechanism of C19and C20accretion products from α-pinene oxidation, which are believed to be efficient SOA precursors. Measurements have been performed in a free-jet flow system. Detection of RO2radicals and accretion products was carried out by recent mass spectrometric techniques using different ionization schemes. Observed C10-RO2radicals from α-pinene ozonolysis were O,O-C10H15(O2)xO2with x= 0, 1, 2, 3 and from the OH radical reaction HO-C10H16(O2)αO2with α = 0, 1, 2. All detected C20accretion products can be explained via the accretion reaction RO2+ R′O2→ ROOR′ + O2starting from the measured C10-RO2radicals. We speculate that C19accretion products are formed in an analogous way assuming CH2O elimination. Addition of isoprene (C5H8), producing C5-RO2radicals, leads to C15accretion products formed via cross-reactions with C10-RO2radicals. This process is competing with the formation of C19/C20products from the pure α-pinene oxidation. A similar behavior has been observed for ethylene additives that form C12accretion products. In the atmosphere, a complex accretion product spectrum from self- and cross-reactions of available RO2radicals can be expected. Modeling atmospheric conditions revealed that C19/C20product formation is only reduced by a factor of 1.2 or 3.6 in isoprene-dominated environments assuming a 2- or 15-fold isoprene concentration over α-pinene, respectively, as present in different forested areas.

The integration of visual stimuli and motor feedback is critical for successful visually guided navigation. These signals have been shown to shape neuronal activity in the primary visual cortex (V1), in an experience-dependent manner. Here, we examined whether visual, reward, and self-motion-related inputs are integrated in order to encode behaviorally relevant locations in V1 neurons. Using a behavioral task in a virtual environment, we monitored layer 2/3 neuronal activity as mice learned to locate a reward along a linear corridor. With learning, a subset of neurons became responsive to the expected reward location. Without a visual cue to the reward location, both behavioral and neuronal responses relied on self-motion-derived estimations. However, when visual cues were available, both neuronal and behavioral responses were driven by visual information. Therefore, a population of V1 neurons encode behaviorally relevant spatial locations, based on either visual cues or on self-motion feedback when visual cues are absent.

Spatial learning requires estimates of location that may be obtained by path integration or from positional cues. Grid and other spatial firing patterns of neurons in the superficial medial entorhinal cortex (MEC) suggest roles in behavioral estimation of location. However, distinguishing the contributions of path integration and cue-based signals to spatial behaviors is challenging, and the roles of identified MEC neurons are unclear. We use virtual reality to dissociate linear path integration from other strategies for behavioral estimation of location. We find that mice learn to path integrate using motor-related self-motion signals, with accuracy that decreases steeply as a function of distance. We show that inactivation of stellate cells in superficial MEC impairs spatial learning in virtual reality and in a real world object location recognition task. Our results quantify contributions of path integration to behavior and corroborate key predictions of models in which stellate cells contribute to location estimation.

Active turbulence generators were incorporated into a wind tunnel to investigate more realistic inflow conditions for a film cooling test rig. The flow field signals are sampled numerically by probes in large eddy simulations (LES) and experimentally using 1D hot-wire measurements to determine turbulence quantities. The LES shows that the turbulence is anisotropic which cannot be detected by the 1D hot-wire. Furthermore, the integral length scale which shall provide insight into the sizes of the turbulent eddies is determined using two approaches. The first uses the one probe at two times correlation method and could be evaluated from the numerical and experimental probes. The second correlation method exploits the spatial resolution in the LES domain using the two probes at one time approach. Both methods show combustor-like turbulence length scales downstream of the active turbulence generators if the triple decomposition is applied onto the velocity signal.

Plant cells use sugars mainly as a source or store of energy and carbon skeletons for anabolic reactions and for osmotic regulation. The perception of sugars and their responses are rather complex including the heterotrimeric G protein pathway and a seven-transmembrane RGS molecule. Previously, we found that endocytosis of the 7TM-RGS leads to sustained activation of the G protein pathway in the genetic model Arabidopsis. Here we show that other plants possess similar endocytosis systems of the 7TM-RGS proteins. A phosphorylation site essential for the endocytosis is well conserved in land plant 7TM-RGS proteins. In addition, conifer and tobacco 7TM-RGS proteins are internalized in response to sugar. These results indicate a universal mechanism to activate G signaling by endocytosis in plant cells that have 7TM-RGS proteins.