Search

Background Imipenem combined with the β-lactamase inhibitor relebactam has broad antibacterial activity, including against carbapenem-resistant gram-negative pathogens. We evaluated efficacy and safety of imipenem/cilastatin/relebactam in treating hospital-acquired/ventilator-associated bacterial pneumonia (HABP/VABP). Methods This was a randomized, controlled, double-blind phase 3 trial. Adults with HABP/VABP were randomized 1:1 to imipenem/cilastatin/relebactam 500 mg/500 mg/250 mg or piperacillin/tazobactam 4 g/500 mg, intravenously every 6 hours for 7–14 days. The primary endpoint was day 28 all-cause mortality in the modified intent-to-treat (MITT) population (patients who received study therapy, excluding those with only gram-positive cocci at baseline). The key secondary endpoint was clinical response 7–14 days after completing therapy in the MITT population. Results Of 537 randomized patients (from 113 hospitals in 27 countries), the MITT population comprised 264 imipenem/cilastatin/relebactam and 267 piperacillin/tazobactam patients; 48.6% had ventilated HABP/VABP, 47.5% APACHE II score ≥15, 24.7% moderate/severe renal impairment, 42.9% were ≥65 years old, and 66.1% were in the intensive care unit. The most common baseline pathogens were Klebsiella pneumoniae (25.6%) and Pseudomonas aeruginosa (18.9%). Imipenem/cilastatin/relebactam was noninferior (P, Imipenem/cilastatin/relebactam was noninferior to piperacillin/tazobactam for treating hospital-acquired and ventilator-associated bacterial pneumonia (HABP/VABP), with a comparable tolerability profile. Imipenem/cilastatin/relebactam is an efficacious treatment option for nosocomial pneumonia, including HABP/VABP in mechanically ventilated and critically ill, high-risk patients.

The Guinier law for magnetic SANS on bulk ferromagnets is introduced and applied to the analysis of nanocrystalline cobalt. The magnetic-field-dependent Guinier radius reflects the characteristic microstructural size and depends on the magnetic interactions., Small-angle scattering of X-rays and neutrons is a routine method for the determination of nanoparticle sizes. The so-called Guinier law represents the low-q approximation for the small-angle scattering curve from an assembly of particles. The Guinier law has originally been derived for nonmagnetic particle-matrix-type systems and it is successfully employed for the estimation of particle sizes in various scientific domains (e.g. soft-matter physics, biology, colloidal chemistry, materials science). An important prerequisite for it to apply is the presence of a discontinuous interface separating particles and matrix. Here, the Guinier law is introduced for the case of magnetic small-angle neutron scattering and its applicability is experimentally demonstrated for the example of nanocrystalline cobalt. It is well known that the magnetic microstructure of nanocrystalline ferromagnets is highly nonuniform on the nanometre length scale and characterized by a spectrum of continuously varying long-wavelength magnetization fluctuations, i.e. these systems do not manifest sharp interfaces in their magnetization profile. The magnetic Guinier radius depends on the applied magnetic field, on the magnetic interactions (exchange, magnetostatics) and on the magnetic anisotropy-field radius, which characterizes the size over which the magnetic anisotropy field is coherently aligned into the same direction. In contrast to the nonmagnetic conventional Guinier law, the magnetic version can be applied to fully dense random-anisotropy-type ferromagnets.

The RTgill-W1 (gill), RTG-2 (gonad), and RTL-W1 (liver) cell lines derived from a freshwater fish rainbow trout (Oncorhynchus mykiss), were used to assess the toxicity of polyethylene terephthalate (PET) and two forms of polyvinyl chloride (PVC). Two size fractions (25-μm and 90-μm particles) were tested for all materials. The highest tested concentration was 1 mg/ml, corresponding to from 70 000 ± 9000 to 620 000 ± 57 000 particles/ml for 25-μm particles and from 2300 ± 100 to 11 000 ± 1000 particles/ml for 90-μm particles (depending on the material). Toxicity differences between commercial PVC dry blend powder and secondary microplastics created from a processed PVC were newly described. After a 24-h exposure, the cells were analyzed for changes in viability, 7-ethoxyresorufin-O-deethylase (EROD) activity, and reactive oxygen species (ROS) generation. In addition to the microplastic suspensions, leachates and particles remaining after leaching resuspended in fresh exposure medium were tested. The particles were subjected to leaching for 1, 8, and 15 days. The PVC dry blend (25 μm and 90 μm) and processed PVC (25 μm) increased ROS generation, to which leached chemicals appeared to be the major contributor. PVC dry blend caused substantially higher ROS induction than processed PVC, showing that the former is not suitable for toxicity testing, as it can produce different results from those of secondary PVC. The 90-μm PVC dry blend increased ROS generation only after prolonged leaching. PET did not induce any changes in ROS generation, and none of the tested polymers had any effect on viability or EROD activity. The importance of choosing realistic extraction procedures for microplastic toxicity experiments was emphasized. Conducting long-term experiments is crucial to detect possible environmentally relevant effects. In conclusion, the tested materials showed no acute toxicity to the cell lines.

Unsupervised cross-lingual pretraining has achieved strong results in neural machine translation (NMT), by drastically reducing the need for large parallel data. Most approaches adapt masked-language modeling (MLM) to sequence-to-sequence architectures, by masking parts of the input and reconstructing them in the decoder. In this work, we systematically compare masking with alternative objectives that produce inputs resembling real (full) sentences, by reordering and replacing words based on their context. We pretrain models with different methods on English$\leftrightarrow$German, English$\leftrightarrow$Nepali and English$\leftrightarrow$Sinhala monolingual data, and evaluate them on NMT. In (semi-) supervised NMT, varying the pretraining objective leads to surprisingly small differences in the finetuned performance, whereas unsupervised NMT is much more sensitive to it. To understand these results, we thoroughly study the pretrained models using a series of probes and verify that they encode and use information in different ways. We conclude that finetuning on parallel data is mostly sensitive to few properties that are shared by most models, such as a strong decoder, in contrast to unsupervised NMT that also requires models with strong cross-lingual abilities., Findings of ACL 2021

Magnetic-field-dependent unpolarized small-angle neutron scattering (SANS) experiments demonstrate that high-pressure torsion (HPT) straining induces spin misalignments in pure Ni, which persist in magnetic fields up to 4 T. The spin-misalignment scattering patterns are elongated perpendicular to the applied magnetic field due to an unusual predominant longitudinal $sin^2(\theta)$-type angular anisotropy. Such a contribution cannot be explained by the conventional second order (in spin misalignment amplitude) micromagnetic SANS theory in the approach-to-saturation regime, nor can its magnitude relative to the other features of the cross sections by the third order micromagnetic SANS theory. This indicates that the high-density of crystal defects induced via HPT straining in Ni makes such higher-order effects in the micromagnetic SANS cross sections observable., Comment: 9 figures

Document-level contextual information has shown benefits to text-based machine translation, but whether and how context helps end-to-end (E2E) speech translation (ST) is still under-studied. We fill this gap through extensive experiments using a simple concatenation-based context-aware ST model, paired with adaptive feature selection on speech encodings for computational efficiency. We investigate several decoding approaches, and introduce in-model ensemble decoding which jointly performs document- and sentence-level translation using the same model. Our results on the MuST-C benchmark with Transformer demonstrate the effectiveness of context to E2E ST. Compared to sentence-level ST, context-aware ST obtains better translation quality (+0.18-2.61 BLEU), improves pronoun and homophone translation, shows better robustness to (artificial) audio segmentation errors, and reduces latency and flicker to deliver higher quality for simultaneous translation.

We report the results of an unpolarized small-angle neutron scattering (SANS) study on Mn-Bi-based rare-earth-free permanent magnets. The magnetic SANS cross section is dominated by long-wavelength transversal magnetization fluctuations, and has been analyzed in terms of the Guinier-Porod model and the distance distribution function. This provides the radius of gyration which, in the remanent state, ranges between about $220\ensuremath{-}240\phantom{\rule{0.16em}{0ex}}\mathrm{nm}$ for the three different alloy compositions investigated. Moreover, computation of the distance distribution function, in conjunction with results for the so-called $s$ parameter obtained from the Guinier-Porod model, indicates that the magnetic scattering of a ${\mathrm{Mn}}_{45}{\mathrm{Bi}}_{55}$ sample has its origin in slightly shape-anisotropic structures.

Sequence-to-sequence models usually transfer all encoder outputs to the decoder for generation. In this work, by contrast, we hypothesize that these encoder outputs can be compressed to shorten the sequence delivered for decoding. We take Transformer as the test bed and introduce a layer of stochastic gates in-between the encoder and the decoder. The gates are regularized using the expected value of the sparsity-inducing L0 penalty, resulting in completely masking-out a subset of encoder outputs. In other words, via joint training, the L0DROP layer forces Transformer to route information through a subset of its encoder states. We investigate the effects of this sparsification on two machine translation and two summarization tasks. Experiments show that, depending on the task, around 40–70% of source encodings can be pruned without significantly compromising quality. The decrease of the output length endows L0DROP with the potential of improving decoding efficiency, where it yields a speedup of up to 1.65× on document summarization and 1.20× on character-based machine translation against the standard Transformer. We analyze the L0DROP behaviourand observe that it exhibits systematic preferences for pruning certain word types, e.g., function words and punctuation get pruned most. Inspired by these observations, we explore the feasibility of specifying rule-based patterns that mask out encoder outputs based on information such as part-of-speech tags, word frequency and word position.

The importance of building semantic parsers which can be applied to new domains and generate programs unseen at training has long been acknowledged, and datasets testing out-of-domain performance are becoming increasingly available. However, little or no attention has been devoted to learning algorithms or objectives which promote domain generalization, with virtually all existing approaches relying on standard supervised learning. In this work, we use a meta-learning framework which targets zero-shot domain generalization for semantic parsing. We apply a model-agnostic training algorithm that simulates zero-shot parsing by constructing virtual train and test sets from disjoint domains. The learning objective capitalizes on the intuition that gradient steps that improve source-domain performance should also improve target-domain performance, thus encouraging a parser to generalize to unseen target domains. Experimental results on the (English) Spider and Chinese Spider datasets show that the meta-learning objective significantly boosts the performance of a baseline parser., NAACL2021 Camera Ready

Objective: Evaluating a solution-oriented clinical supervision to improve the quality of care. Design: We performed a randomized, longitudinal controlled trial. A total of 32 health districts were involved in the study. For the evaluation of the intervention, the following dimensions were collected as indicators of the quality of supervision: Quality of life (FACT-G, SF12, POS), psychological stress, depression, burn-out (HADS, BDI-II, VAS scales, HPS), sense of coherence (SOC-13), satisfaction with care, communication and support from the patients and relatives (VAS scales) and working conditions (COPSOQ) from the nursing staff and family doctors. Results: Of the 85 subscales, the SOC Nursing Sum Score (p=0.017), the SF-12 Nursing Sum Scale (p=0.036), and the COPSOQ Scales of General Practitioners showed significant differences in developmental opportunities (p=0.020), leadership (p=0.003), social support (p=0.001) and community spirit (p=0.024). At the second point time of the study, significant differences were found in the subscales of the Palliative Care Outcome Scale (POS) and the subscale of the test instrument Functional Assessment of Cancer Therapy - General (FACT-G) FUNCTIONAL WELL-BEING of the patients. The satisfaction values of nurses and general practitioners with the supervision showed an extremely positive assessment of both nurses and general practitioners regarding supervision. Conclusions: Supervision affects positively the process of palliative home care. It seems important to adjust the number of supervision meetings according to the needs of the individual team in order to achieve an optimized team performance.

Word sense disambiguation is a well-known source of translation errors in NMT. We posit that some of the incorrect disambiguation choices are due to models' over-reliance on dataset artifacts found in training data, specifically superficial word co-occurrences, rather than a deeper understanding of the source text. We introduce a method for the prediction of disambiguation errors based on statistical data properties, demonstrating its effectiveness across several domains and model types. Moreover, we develop a simple adversarial attack strategy that minimally perturbs sentences in order to elicit disambiguation errors to further probe the robustness of translation models. Our findings indicate that disambiguation robustness varies substantially between domains and that different models trained on the same data are vulnerable to different attacks., Comment: Accepted to EMNLP 2020

Information in speech signals is not evenly distributed, making it an additional challenge for end-to-end (E2E) speech translation (ST) to learn to focus on informative features. In this paper, we propose adaptive feature selection (AFS) for encoder-decoder based E2E ST. We first pre-train an ASR encoder and apply AFS to dynamically estimate the importance of each encoded speech feature to SR. A ST encoder, stacked on top of the ASR encoder, then receives the filtered features from the (frozen) ASR encoder. We take L0DROP (Zhang et al., 2020) as the backbone for AFS, and adapt it to sparsify speech features with respect to both temporal and feature dimensions. Results on LibriSpeech En-Fr and MuST-C benchmarks show that AFS facilitates learning of ST by pruning out ~84% temporal features, yielding an average translation gain of ~1.3-1.6 BLEU and a decoding speedup of ~1.4x. In particular, AFS reduces the performance gap compared to the cascade baseline, and outperforms it on LibriSpeech En-Fr with a BLEU score of 18.56 (without data augmentation), Comment: EMNLP2020 Findings; source code is at https://github.com/bzhangGo/zero

The interaction of microplastics (MPs) and common environmental organic pollutants has been a frequently discussed topic in recent years. Although the estimated contamination caused by MPs in terrestrial ecosystems is one order of magnitude higher than that in the oceans, experiments have been conducted solely in an aqueous matrix. Therefore, an experiment was carried out with two soils differing in their concentrations of polycyclic aromatic hydrocarbons (PAHs) and polyurethane foams used for scent fences along roads and crop fields. Two types of polyurethane foam (biodegradable and conventional in aged and unaged form) were exposed to soils containing PAHs that originated from historically contaminated localities. The exposure lasted 28 days, and a newly developed three-step procedure to separate MPs from soil was then applied. Biodegradable polyurethane MPs exhibited a strong tendency to accumulate PAHs after 7 days, and their concentrations significantly grew over time. In contrast, the sorption of PAHs on conventional polyurethane MPs was substantially lower (a maximum of 3.6 times higher concentration than that in the soil). Neither type of foam changed their sorption behaviors after the aging procedure. The results indicate that the flexibility of the polyurethane polymeric network could be the main driving factor for the sorption.

By means of temperature and wavelength-dependent small-angle neutron scattering (SANS) experiments on sintered isotropic and textured Nd-Fe-B magnets we provide evidence for the existence of an anisometric structure in the microstructure of the textured magnets. This conclusion is reached by observing a characteristic cross-shaped angular anisotropy in the total unpolarized SANS cross section at temperatures well above the Curie temperature. Comparison of the experimental SANS data to a microstructural model based on the superquadrics form factor allows us to estimate the shape and lower bounds for the size of the structure. Subtraction of the scattering cross section in the paramagnetic regime from data taken at room temperature provides the magnetic SANS cross section. Surprisingly, the anisotropy of the magnetic scattering is very similar to the nuclear SANS signal, suggesting that the nuclear structure is decorated by the magnetic moments via spin-orbit coupling. Based on the computation of the two-dimensional correlation function we estimate lower bounds for the longitudinal and transversal magnetic correlation lengths.

Massively multilingual models for neural machine translation (NMT) are theoretically attractive, but often underperform bilingual models and deliver poor zero-shot translations. In this paper, we explore ways to improve them. We argue that multilingual NMT requires stronger modeling capacity to support language pairs with varying typological characteristics, and overcome this bottleneck via language-specific components and deepening NMT architectures. We identify the off-target translation issue (i.e. translating into a wrong target language) as the major source of the inferior zero-shot performance, and propose random online backtranslation to enforce the translation of unseen training language pairs. Experiments on OPUS-100 (a novel multilingual dataset with 100 languages) show that our approach substantially narrows the performance gap with bilingual models in both one-to-many and many-to-many settings, and improves zero-shot performance by ~10 BLEU, approaching conventional pivot-based methods., Comment: ACL2020

Opinion summarization is the task of automatically creating summaries that reflect subjective information expressed in multiple documents, such as product reviews. While the majority of previous work has focused on the extractive setting, i.e., selecting fragments from input reviews to produce a summary, we let the model generate novel sentences and hence produce abstractive summaries. Recent progress in summarization has seen the development of supervised models which rely on large quantities of document-summary pairs. Since such training data is expensive to acquire, we instead consider the unsupervised setting, in other words, we do not use any summaries in training. We define a generative model for a review collection which capitalizes on the intuition that when generating a new review given a set of other reviews of a product, we should be able to control the "amount of novelty" going into the new review or, equivalently, vary the extent to which it deviates from the input. At test time, when generating summaries, we force the novelty to be minimal, and produce a text reflecting consensus opinions. We capture this intuition by defining a hierarchical variational autoencoder model. Both individual reviews and the products they correspond to are associated with stochastic latent codes, and the review generator ("decoder") has direct access to the text of input reviews through the pointer-generator mechanism. Experiments on Amazon and Yelp datasets, show that setting at test time the review's latent code to its mean, allows the model to produce fluent and coherent summaries reflecting common opinions., ACL 2020

This paper presents the characterization of the intermetallic phases of laser-welded Al-Cu joints by synchrotron trough-the-substrate microdiffraction combined with SEM/EDX and etched/non-etched optical micrographs. Transmission microdiffraction offers the spatial resolution and beam flux necessary to study the intermetallic phases, which in conjunction with the excellent contrast of etched micrographs, rendered possible the phase identification. It was found that the major phases are Al2Cu (θ), Al4Cu9 (γ1) and AlCu (η2) were formed. The Al3Cu4 (ζ1) and δ phases where formed in less amount and found to be the primary cause for cracks inside the weld seam and therefore for the failure of the joint.

Hot-deformed anisotropic $\mathrm{Nd}\ensuremath{-}\mathrm{Fe}\ensuremath{-}\mathrm{B}$ nanocrystalline magnets have been subjected to the grain-boundary diffusion process (GBDP) using a ${\mathrm{Pr}}_{70}{\mathrm{Cu}}_{30}$ eutectic alloy. The resulting grain microstructure, consisting of shape-anisotropic $\mathrm{Nd}\ensuremath{-}\mathrm{Fe}\ensuremath{-}\mathrm{B}$ nanocrystals surrounded by a $\mathrm{Pr}\ensuremath{-}\mathrm{Cu}$-rich intergranular grain-boundary phase, has been investigated using unpolarized small-angle neutron scattering and very small-angle neutron scattering. The neutron data have been analyzed using the generalized Guinier-Porod model and by computing, model independently, the distance distribution function. We find that the GBDP results in a change of the geometry of the scattering particles: In the small-$q$ regime, the scattering from the as-prepared sample exhibits a slope of about 2, which is characteristic for the scattering from two-dimensional platelet-shaped objects, while the GBDP sample manifests a slope of about 1, which is the scattering signature of one-dimensional elongated objects. The evolution of the Porod exponent indicates the smoothing of the grain surfaces due to the GBDP, which is accompanied by an increase of the coercivity.

The transformer is a state-of-the-art neural translation model that uses attention to iteratively refine lexical representations with information drawn from the surrounding context. Lexical features are fed into the first layer and propagated through a deep network of hidden layers. We argue that the need to represent and propagate lexical features in each layer limits the model's capacity for learning and representing other information relevant to the task. To alleviate this bottleneck, we introduce gated shortcut connections between the embedding layer and each subsequent layer within the encoder and decoder. This enables the model to access relevant lexical content dynamically, without expending limited resources on storing it within intermediate states. We show that the proposed modification yields consistent improvements over a baseline transformer on standard WMT translation tasks in 5 translation directions (0.9 BLEU on average) and reduces the amount of lexical information passed along the hidden layers. We furthermore evaluate different ways to integrate lexical connections into the transformer architecture and present ablation experiments exploring the effect of proposed shortcuts on model behavior., Comment: Accepted submission to WMT 2019

Shell ferromagnetism is a new functional property of certain Heusler alloys which was recently observed in ${\mathrm{Ni}}_{50}{\mathrm{Mn}}_{45}{\mathrm{In}}_{5}$. We report the results of a comparative study of the magnetic microstructure of bulk ${\mathrm{Ni}}_{50}{\mathrm{Mn}}_{45}{\mathrm{In}}_{5}$ Heusler alloys using magnetometry, synchrotron x-ray diffraction, and magnetic small-angle neutron scattering (SANS). By combining unpolarized and spin-polarized SANS (so-called POLARIS) we demonstrate that a number of important conclusions regarding the mesoscopic spin structure can be made. In particular, the analysis of the magnetic neutron data suggests that nanoprecipitates with an effective ferromagnetic component form in an antiferromagnetic matrix on field annealing at $700\phantom{\rule{0.16em}{0ex}}\mathrm{K}$. These particles represent sources of perturbation, which seem to give rise to magnetically disordered regions in the vicinity of the particle-matrix interface. Analysis of the spin-flip SANS cross section via the computation of the correlation function yields a value of $\ensuremath{\sim}55$ nm for the particle size and $\ensuremath{\sim}20$ nm for the size of the spin-canted region.

The antisymmetric Dzyaloshinskii-Moriya interaction (DMI) plays a decisive role for the stabilization and control of chirality of skyrmion textures in various magnetic systems exhibiting a noncentrosymmetric crystal structure. A less studied aspect of the DMI is that this interaction is believed to be operative in the vicinity of lattice imperfections in crystalline magnetic materials, due to the local structural inversion symmetry breaking. If this scenario leads to an effect of sizable magnitude, it implies that the DMI introduces chirality into a very large class of magnetic materials---defect-rich systems such as polycrystalline magnets. Here, we show experimentally that the microstructural-defect-induced DMI gives rise to a polarization-dependent asymmetric term in the small-angle neutron scattering (SANS) cross section of polycrystalline ferromagnets with a centrosymmetric crystal structure. The results are supported by theoretical predictions using the continuum theory of micromagnetics. This effect, conjectured already by Arrott in 1963, is demonstrated for nanocrystalline terbium and holmium (with a large grain-boundary density), and for mechanically-deformed microcrystalline cobalt (with a large dislocation density). Analysis of the scattering asymmetry allows one to determine the defect-induced DMI constant, $D = 0.45 \pm 0.07 \, \mathrm{mJ/m^2}$ for Tb at $100 \, \mathrm{K}$. Our study proves the generic relevance of the DMI for the magnetic microstructure of defect-rich ferromagnets with vanishing intrinsic DMI. Polarized SANS is decisive for disclosing the signature of the defect-induced DMI, which is related to the unique dependence of the polarized SANS cross section on the chiral interactions. The findings open up the way to study defect-induced skyrmionic magnetization textures in disordered materials.

The goal of homomorphic encryption is to encrypt data such that another party can operate on it without being explicitly exposed to the content of the original data. We introduce an idea for a privacy-preserving transformation on natural language data, inspired by homomorphic encryption. Our primary tool is {\em obfuscation}, relying on the properties of natural language. Specifically, a given English text is obfuscated using a neural model that aims to preserve the syntactic relationships of the original sentence so that the obfuscated sentence can be parsed instead of the original one. The model works at the word level, and learns to obfuscate each word separately by changing it into a new word that has a similar syntactic role. The text obfuscated by our model leads to better performance on three syntactic parsers (two dependency and one constituency parsers) in comparison to an upper-bound random substitution baseline. More specifically, the results demonstrate that as more terms are obfuscated (by their part of speech), the substitution upper bound significantly degrades, while the neural model maintains a relatively high performing parser. All of this is done without much sacrifice of privacy compared to the random substitution upper bound. We also further analyze the results, and discover that the substituted words have similar syntactic properties, but different semantic content, compared to the original words., Comment: Accepted to IWPT 2020

Isotropic Nd Fe B-based sintered magnets, before and after a (Tb-doped) grain-boundary diffusion process, have been investigated by means of magnetic-field-dependent small-angle neutron scattering. Compared to the untreated sample, we found a reduced correlation length in the grain boundary diffused sample which, in an applied-field range of 8.5 T, varies between 30 and 35 nm – about 15% smaller than in the untreated specimen. This observation is related to the increased magnetic anisotropy field of the nucleation sites for magnetization reversal, and may be explained by a reduction in the effective defect size responsible for the magnetic inhomogeneities.

Standard machine translation systems process sentences in isolation and hence ignore extra-sentential information, even though extended context can both prevent mistakes in ambiguous cases and improve translation coherence. We introduce a context-aware neural machine translation model designed in such way that the flow of information from the extended context to the translation model can be controlled and analyzed. We experiment with an English-Russian subtitles dataset, and observe that much of what is captured by our model deals with improving pronoun translation. We measure correspondences between induced attention distributions and coreference relations and observe that the model implicitly captures anaphora. It is consistent with gains for sentences where pronouns need to be gendered in translation. Beside improvements in anaphoric cases, the model also improves in overall BLEU, both over its context-agnostic version (+0.7) and over simple concatenation of the context and source sentences (+0.6).

Background: Transfusion-related ARDS is a potentially fatal complication associated with transfusion of blood products. Adequate respiratory support of patient is crucial to its management. The role of NIV has been largely controversial due to the lack of evidence. This study aims at justifying the utility of NIV in management of TRALI. Materials and Methods: A retrospective case study of ICU transfused patients during 2010-2016 has showed 17(13 male, 4 female) patients with ARDS. Respiratory support involved NIV (CPAP), its efficacy determined by SpO2, PaO2/FiO2, X-ray. Results: Data analysis showed that each patient has been transfused at least 1000 ml of FFP during 6-12h. 4 patients (23.5%) developed hypoxemia within 6h after transfusion and 13 patients (76.5%) in 6-24h. Bilateral infiltration on frotal X-ray followed in 24-48h after hypoxemic changes. The majority of patients (88.2%) was diagnosed with mild ARDS, the rest– moderate ARDS. 3 patients have been mechanically ventilated due to postoperative concerns. 14 patients have been on spontaneous ventilation and required NIV due to progressive hypoxemia. Before NIV the median PaO2/FiO2 ratio was 235 mm Hg, 6 hours after- 310, a significant increase(p=0,0009; 95 %).In 12 (85,7%) out of 14 patients there was no need for invasive ventilation. The clinical picture of ARDS has resolved by 5 days. Conclusions: 1. The onset of transfusion-related ARDS was mostly observed in > 6 hours after transfusion. There was also a reported delay of X-ray changes if compared with hypoxemia. 2. As dynamics suggests, selective use of NIV with mild transfusion-related ARDS results in improved oxygenation and reduced incidence of invasive ventilation. References: J.E Levitt et al. Expert Rev Respir Med.2010;4(6):785–797

Coreference evaluation metrics are hard to optimize directly as they are non-differentiable functions, not easily decomposable into elementary decisions. Consequently, most approaches optimize objectives only indirectly related to the end goal, resulting in suboptimal performance. Instead, we propose a differentiable relaxation that lends itself to gradient-based optimisation, thus bypassing the need for reinforcement learning or heuristic modification of cross-entropy. We show that by modifying the training objective of a competitive neural coreference system, we obtain a substantial gain in performance. This suggests that our approach can be regarded as a viable alternative to using reinforcement learning or more computationally expensive imitation learning., Comment: 10 pages. CoNLL

Recent research in psycholinguistics has provided increasing evidence that humans predict upcoming content. Prediction also affects perception and might be a key to robustness in human language processing. In this paper, we investigate the factors that affect human prediction by building a computational model that can predict upcoming discourse referents based on linguistic knowledge alone vs. linguistic knowledge jointly with common-sense knowledge in the form of scripts. We find that script knowledge significantly improves model estimates of human predictions. In a second study, we test the highly controversial hypothesis that predictability influences referring expression type but do not find evidence for such an effect., Comment: 14 pages, published at TACL, 2017, Volume-5, Pg 31-44, 2017

We introduce a simple and accurate neural model for dependency-based semantic role labeling. Our model predicts predicate-argument dependencies relying on states of a bidirectional LSTM encoder. The semantic role labeler achieves competitive performance on English, even without any kind of syntactic information and only using local inference. However, when automatically predicted part-of-speech tags are provided as input, it substantially outperforms all previous local models and approaches the best reported results on the English CoNLL-2009 dataset. We also consider Chinese, Czech and Spanish where our approach also achieves competitive results. Syntactic parsers are unreliable on out-of-domain data, so standard (i.e., syntactically-informed) SRL models are hindered when tested in this setting. Our syntax-agnostic model appears more robust, resulting in the best reported results on standard out-of-domain test sets., To appear in CoNLL 2017

Semantic role labeling (SRL) is the task of identifying the predicate-argument structure of a sentence. It is typically regarded as an important step in the standard NLP pipeline. As the semantic representations are closely related to syntactic ones, we exploit syntactic information in our model. We propose a version of graph convolutional networks (GCNs), a recent class of neural networks operating on graphs, suited to model syntactic dependency graphs. GCNs over syntactic dependency trees are used as sentence encoders, producing latent feature representations of words in a sentence. We observe that GCN layers are complementary to LSTM ones: when we stack both GCN and LSTM layers, we obtain a substantial improvement over an already state-of-the-art LSTM SRL model, resulting in the best reported scores on the standard benchmark (CoNLL-2009) both for Chinese and English., To appear in EMNLP 2017

Information-theoretic measures are among the most standard techniques for evaluation of clustering methods including word sense induction (WSI) systems. Such measures rely on sample-based estimates of the entropy. However, the standard maximum likelihood estimates of the entropy are heavily biased with the bias dependent on, among other things, the number of clusters and the sample size. This makes the measures unreliable and unfair when the number of clusters produced by different systems vary and the sample size is not exceedingly large. This corresponds exactly to the setting of WSI evaluation where a ground-truth cluster sense number arguably does not exist and the standard evaluation scenarios use a small number of instances of each word to compute the score. We describe more accurate entropy estimators and analyze their performance both in simulations and on evaluation of WSI systems.

The article presents the results of combination treatment of patients with acute ischemic stroke (the NIH Stroke Scale – 12.04±0.57). The efficacy of conventional therapy (the control group) was compared with treatment regimen using amantadine sulfate (PK-Merz) (the main group). In patients undergoing combination treatment lost functions were restored quite promptly and 2 months after the observation their functional state was satisfactory (the NIH Stroke Scale: the main group – 2.49±0.78, the control group – 5.53±0.69, p=0.009); moreover, the subscale “language” differed significantly from that in patients receiving basic therapy only (р

We report the results of a comparative study of the magnetic microstructure of textured and isotropic $\mathrm{Nd}_2\mathrm{Fe}_{14}\mathrm{B}/\alpha$-$\mathrm{Fe}$ nanocomposites using magnetometry, transmission electron microscopy, synchrotron x-ray diffraction, and, in particular, magnetic small-angle neutron scattering (SANS). Analysis of the magnetic neutron data of the textured specimen and computation of the correlation function of the spin misalignment SANS cross section suggests the existence of inhomogeneously magnetized regions on an intraparticle nanometer length scale, about $40-50 \, \mathrm{nm}$ in the remanent state. Possible origins for this spin disorder are discussed: it may originate in thin grain-boundary layers (where the materials parameters are different than in the $\mathrm{Nd}_2\mathrm{Fe}_{14}\mathrm{B}$ grains), or it may reflect the presence of crystal defects (introduced via hot pressing), or the dispersion in the orientation distribution of the magnetocrystalline anisotropy axes of the $\mathrm{Nd}_2\mathrm{Fe}_{14}\mathrm{B}$ grains. X-ray powder diffraction data reveal a crystallographic texture in the direction perpendicular to the pressing direction -- a finding which might be related to the presence of a texture in the magnetization distribution, as inferred from the magnetic SANS data.

We have investigated the effect of the annealing conditions (heating rate and temperature) on the magnetic microstructure of sintered Nd-Fe-B magnets by means of magnetometry, scanning electron microscopy, high-energy synchrotron x-ray diffraction, and small-angle neutron scattering (SANS). While the temperature treatment has a strong effect on the coercivity (reduction by about 50% on annealing), the associated changes in the microstructure do surprisingly not show up (or at best only very weakly) in the neutron-scattering signal, which probes a mesoscopic real-space length scale ranging between about 1–300 nm. On the other hand, the x-ray data reveal microstructural changes in the Nd-rich phases, presumably due to modifications in grain-boundary regions. Moreover, we observe an unusual diamond-shaped angular anisotropy in the SANS cross section, which strongly points towards the existence of texture in the nuclear microstructure.

Tsarichanskoe is one of the largest oil fields in the Volga-Ural region of Russia with proven reserves oven 25 million tons of oil equivalent. An ambitious goal to produce up to 2 million tons over the next 5 years was set to achieve. The target formation to produce from is a Devonian reservoir also known as DKT. It is represented by alternation of sandstone, mudstone and siltstone and has challenging properties to deal with such as low permeability and low temperature (60–70C). Economic production from these reservoirs requires an effective stimulation. Conventional hydraulic fracturing in vertical and horizontal wells has been intensively utilized in the field with varying success since the beginning of its development. In some cases, wells completed with conventional fracturing underperform expectations and demonstrate rapid production decline. In other, complex geological conditions restrict treatment size with fracture conductivity being compromised. Therefore, the need exists for a novel stimulation technique capable to perform in the given environment and unlock the reservoir to its full potential. Channel fracturing technique has been successfully deployed in Russia with more than 170 treatments pumped in different fields. The channel fracturing technique creates highly conductive open pathways within the proppant pack by combining specific pumping protocols with fluid and fiber technologies. It is aimed at maximizing fracture conductivity and establishes unrestricted flow of fluids from the reservoir to the wellbore through the network for channels formed by heterogeneous proppant placement. To ensure stability of the channels network, the use of appropriate degradable fiber materials is imperative. The fiber additive has to be chosen based upon specific reservoir temperature. Several treatments were pumped with unique low-temperature fiber among the first trials worldwide. Successful application of low-temperature fiber enabled channel fracturing technology for Tsarichanskoe. Field campaign was supported by extensive laboratory analysis of the fiber to prove its mechanical properties and required functionality for channel fracturing in low temperature reservoirs. Initial production analysis indicates that application of channel fracturing in conjunction with the unique degradable fiber resulted in a sustainable oil rate increase up to 30% in comparison with conventional fracturing. The step change in production was achieved with noticeable reduction in proppant and water consumption by 40% and 15% respectively. Successful deployment of the channel fracturing in Tsarichanskoe field offers new opportunities for effective stimulation of low temperature formations in Russia.

We introduce a new approach to unsupervised estimation of feature-rich semantic role labeling models. Our model consists of two components: (1) an encoding component: a semantic role labeling model which predicts roles given a rich set of syntactic and lexical features; (2) a reconstruction component: a tensor factorization model which relies on roles to predict argument fillers. When the components are estimated jointly to minimize errors in argument reconstruction, the induced roles largely correspond to roles defined in annotated resources. Our method performs on par with most accurate role induction methods on English and German, even though, unlike these previous approaches, we do not incorporate any prior linguistic knowledge about thelanguages.

In this paper, we consider the model of server traffic when the traffic is separated into several streams. The amount of transferred data differs for different streams. Based on real traffic measurements we proposed the server traffic model where traffic of every stream is described by the same independent processes, but each process has its own time scale. We show that for traffic analysis as well as for developing of the most effective methods of control of this traffic, it is necessary to correctly identify the time scale for each stream, as well as the time scale of traffic fluctuations those have a significant effect to QoS.

Multiple documents describing the same or closely related sets of events are common and often easy to obtain: for example, consider document clusters on a news aggregator site or multiple reviews of the same product or service. Even though each such document discusses a similar set of topics, they provide alternative views or complimentary information on each of these topics. We argue that revealing hidden relations by jointly segmenting the documents, or, equivalently, predicting links between topically related segments in different documents would help to visualize documents of interest and construct friendlier user interfaces. In this paper, we refer to this problem as multi-document topic segmentation. We propose an unsupervised Bayesian model for the considered problem that models both shared and document-specific topics, and utilizes Dirichlet process priors to determine the effective number of topics. We show that topic segmentation can be inferred efficiently using a simple split-merge sampling algorithm. The resulting method outperforms baseline models on four datasets for multi-document topic segmentation.

Motivated by the large number of languages (seven) and the short development time (two months) of the 2009 CoNLL shared task, we exploited latent variables to avoid the costly process of hand-crafted feature engineering, allowing the latent variables to induce features from the data. We took a pre-existing generative latent variable model of joint syntactic-semantic dependency parsing, developed for English, and applied it to six new languages with minimal adjustments. The parser's robustness across languages indicates that this parser has a very general feature set. The parser's high performance indicates that its latent variables succeeded in inducing effective features. This system was ranked third overall with a macro averaged F1 score of 82.14%, only 0.5% worse than the best system.

In this paper we present a novel framework for extracting the ratable aspects of objects from online user reviews. Extracting such aspects is an important challenge in automatically mining product opinions from the web and in generating opinion-based summaries of user reviews [18, 19, 7, 12, 27, 36, 21]. Our models are based on extensions to standard topic modeling methods such as LDA and PLSA to induce multi-grain topics. We argue that multi-grain models are more appropriate for our task since standard models tend to produce topics that correspond to global properties of objects (e.g., the brand of a product type) rather than the aspects of an object that tend to be rated by a user. The models we present not only extract ratable aspects, but also cluster them into coherent topics, e.g., 'waitress' and 'bartender' are part of the same topic 'staff' for restaurants. This differentiates it from much of the previous work which extracts aspects through term frequency analysis with minimal clustering. We evaluate the multi-grain models both qualitatively and quantitatively to show that they improve significantly upon standard topic models.