Search

It is hard to evaluate translations objectively and accurately, which limits the applications of machine translation. In this article, we assume that the above phenomenon is caused by noise interference during translation evaluation, and we handle the problem through a perspective of causal inference. We assume that the observable translation score is affected by the unobservable true translation quality and some noise simultaneously. If there is a variable that is related to the noise and independent to the true translation quality, the related noise can be eliminated by removing the effect of that variable from the observed score. Based on the above causality hypothesis, this article studies the length bias problem of beam search for neural machine translation (NMT) and the input related noise problem of translation quality estimation (QE). For the NMT length bias problem, we conduct the experiments on four typical NMT tasks (Uyghur–Chinese, Chinese–English, English–German, and English–French) with different scales of datasets. Comparing with previous approaches, the proposed causal motivated method is model-agnostic and does not require supervised training. For QE tasks, we conduct the experiments on the WMT’20 submissions. Experimental results show that the denoised QE results gain better Pearson’s correlation scores with human assessed scores compared to the original submissions. Further analyses on the NMT and QE tasks also demonstrate the rationality of the empirical assumptions made on our methods.

Neural variational inference-based topic modeling has gained great success in mining abstract topics from documents. However, these topic models usually mainly focus on optimizing the topic proportions for documents, while the quality and the internal construction of topics are usually neglected. Specifically, these models lack the guarantee that semantically related words are supposed to be assigned to the same topic and are difficult to ensure the interpretability of topics. Moreover, many topical words recur frequently in the top words of different topics, which makes the learned topics semantically redundant and similar, and of little significance for further study. To solve the above problems, we propose a novel neural topic model called Neural Variational Gaussian Mixture Topic Model (NVGMTM). We use Gaussian distribution to depict the semantic relevance between words in the topics. Each topic in NVGMTM is considered as a multivariate Gaussian distribution over words in the word-embedding space. Thus, semantically related words share similar probabilities in each topic, which makes the topics more coherent and interpretable. Experimental results on two public corpora show the proposed model outperforms the state-of-the-art baselines.

SUMMARY Due to the strong heterogeneity of the reservoir, it is difficult to extend the classical reservoir parameter prediction model to the new work area. Traditional geological methods rely on experience, while machine learning algorithms lack consideration of reservoir heterogeneity. This makes it impossible for these methods to achieve good predictive performance in new work areas. In this paper, we discuss designing a multilayer deep transfer learning network (MDTL). The network enables accurate prediction of reservoir parameters in new areas based on core and logging data from mature areas. For the architectural design, we establish a network structure that realizes data interaction in different work areas. For network parameter optimization, we design a multilayer maximum mean discrepancy loss function. It guides network training to learn feature knowledge about data in different work areas. For logging data processing, we explore a support vector machine method to remove abnormal noise in logging data. We apply MDTL to predict shale gas porosity and total organic carbon content in southern Sichuan. Compared to state-of-the-art reservoir parameter prediction models, MDTL achieves the best performance in new well areas. The network integrates the idea of multilayer transfer learning, reduces the influence of reservoir heterogeneity and effectively ensures the prediction accuracy.

Marine–continental transitional shale is one of the most promising targets for shale gas exploration in the Lower Yangtze region. To investigate the sedimentary environments and the regularity of the enrichment of the Longtan shale, multiple techniques including core and thin-section observations, geochemical and elemental analyses, X-ray diffraction, scanning electron microscopy (SEM), and low-pressure nitrogen adsorption (LPNA) were used to analyze the sedimentology, mineralogy, and pore structure of the Longtan shale. The core descriptions and thin-section observations showed that the Longtan shale was deposited in marine–delta transitional environments including delta-front, shore swamp, mixed tidal flat and shallow shelf environments. The Sr/Cu, V/Cr, CIA, EF (Mo), EF (U), and other major and trace element results indicated warm and moist climates and water-reducing conditions in the Longtan period. Both the climate and water conditions were favorable for organic matter production and preservation. The geochemical results showed that the Longtan shale was in the overmature stage (Ro values ranging from 2.4% to 3.57%) and that the average total organic carbon (TOC) content was 5.76%. The pore system of the Longtan shale consisted of inorganic pores with a small number of organic pores and microfractures. The porosity and specific surface area were mainly affected by the TOC and clay mineral contents. An effective combination of brittle mineral particles, organic matter, and clay minerals provided the necessary conditions for pore preservation. The organic pores, intergranular pores in clay minerals, and brittle mineral pores formed the main network system for the Longtan shale. In summary, the lithological combinations, organic geochemistry, and pore structure system were all affected by the sedimentary environments.

Paper-based electronics have attracted much attention due to their softness, degradability, and low cost. However, paper-based sensors are difficult to apply to high-humidity environments or even underwater. Here, we report a fully paper-integrated piezoresistive sensing system that exhibits flexibility, waterproofing, air permeability, and biocompatibility. This system consists of hydrophobic paper as the substrate and encapsulation layer, conductive paper with a double ‘zig-zag’ and dotted surface structure as the sensing layer, and silver paste films as the interconnects. The structural design of the sensing layer helps to increase the contact area in adjacent layers under pressure and further improves the pressure sensitivity. The piezoresistive system can be worn on human skin in the ambient environment, wet environment, and water for real-time monitoring of physiological signals with air permeability and waterproofing due to its hydrophobic fiber structure. Such a device provides a reliable, economical, and eco-friendly solution to wearable technologies.

Flexible electronics is an emerging field of research involving multiple disciplines, which include but not limited to physics, chemistry, materials science, electronic engineering, and biology. However, the broad applications of flexible electronics are still restricted due to several limitations, including high Young's modulus, poor biocompatibility, and poor responsiveness. Innovative materials aiming for overcoming these drawbacks and boost its practical application is highly desirable. Hydrogel is a class of 3D crosslinked hydrated polymer networks, and its exceptional material properties render it a promising candidate for the next generation of flexible electronics. Here, we review the latest methods of synthesizing advanced functional hydrogels and the state-of-art applications of hydrogel-based flexible electronics in various fields. More importantly, we discuss on the correlation between properties of the hydrogel and device performance, to have better understanding of the development of flexible electronics by using environmentally responsive hydrogels. Lastly, we provide perspectives on the current challenges and future directions in the development of hydrogel-based multifunctional flexible electronics. This article is protected by copyright. All rights reserved.

Currently, the treatment of osteomyelitis poses a great challenge to clinical orthopedics. The use of biodegradable materials combined with antibiotics provides a completely new option for the treatment of osteomyelitis. In this study, vancomycin hydrochloride (VANCO) loaded poly (lactic-co-glycolic acid) (PLGA) microspheres were prepared by a double emulsion solvent evaporation method, and the in vitro drug release behaviors of the drug loaded microspheres were explored after coating with different concentrations of silk fibroin (SF). Drug loading, encapsulation efficiency, Scanning electron microscopy, particle size analysis, Fourier transform infrared spectroscopy, hydrophilicity, in vitro drug release, and in vitro antibacterial activity were evaluated. The results showed that the drug loading of vancomycin loaded PLGA microspheres was (24.11 ±1.72)%, and the encapsulation efficiency was (48.21 ±3.44)%. The in vitro drug release indicated that the drug loaded microspheres showed an obvious initial burst release, and the drug loaded microspheres coated with SF could alleviate the initial burst release in varying degrees. It also can reduce the amount of cumulative drug release, and the effect of microspheres coated with 0.1% concentration of SF is the best. The time of in vitro drug release in different groups of drug loaded microspheres can be up to 28 days. The microspheres coated with (0.1%SF) or without (0%SF) SF showed a cumulative release of (82.50±3.51)% and (67.70±3.81)%,respectively. Therefore, the surface coating with SF of vancomycin loaded microspheres can alleviate the initial burst release, reduce the cumulative drug release, potentially prolong the drug action time, and improve the anti-infection effect.

Paper-based sensors exploiting the advantages of paper can replace traditional substrate materials for building sensors which are simple to manufacture, inexpensive, easy-to-operate, portable and disposable.

Based on the lithologies, sedimentary structures, graptolite zones, inorganic geochemical characteristics, electrical data of 110 shale gas wells in southern Sichuan Basin and the mineral quantitative analysis technology of scanning electron microscope, the stratigraphic sequences of the Upper Ordovician Katian Stage–Himantian Stage–Silurian Rhuddanian Stage–Aeronian Stage are divided, the sedimentary characteristics and fourth-order sequence evolution are analyzed. The target layer can be divided into two sequences, namely SQ1 and SQ2. According to Ordovician–Silurian sedimentary background, the gamma value of the target layer and U/Th, 5 maximum flooding surfaces and 12 system tracts are identified. According to system tracts and their combinations, eight fourth-order sequences are identified, namely, Pss1–Pss8 from old to new. The development period and scale of dominant shale facies from Katian stage to Aeronian stage in southern Sichuan are restored. The best-quality dolomite/calcite-bearing siliceous shale facies, siliceous shale facies, clay-bearing siliceous shale facies and feldspar-bearing siliceous shale facies mainly occur in Pss3–Pss5 of Weiyuan, Western Chongqing and Luzhou, Pss6 of Western Changning– Northern Luzhou–Central Western Chongqing and Pss3–Pss4 of Changning. The siliceous clay shale facies second in quality mainly occurs in Pss6 of Southern Luzhou–Changning area (excluding Western Changning area), Pss7 of Eastern Weiyuan– Northern Western Chongqing–Southern Luzhou and Pss8 of Northern Luzhou–Weiyuan–Western Chongqing. The fourth-order sequence evolution model of Katian stage–Aeronian stage in southern Sichuan is established. During the depositional period of Pss1–Pss8, the sea level had six regressions and five transgressions, and the first transgression SQ2-MFS1 after glaciation was the largest flooding surface.

Currently, Luzhou in the Sichuan Basin is a focal point for shale-gas exploration and development in China. However, a lack of detailed research on the mineral composition of the Wufeng Formation-Longmaxi Formation (WF-LF) shale is hindering the extraction of deep-buried shale gas in the Luzhou shale play. Herein, a field emission scanning electron microscope (FESEM) equipped with the Advanced Mineral Identification and Characterization System (AMICS) software was employed to analyze the mineral composition of the WF-LF shale from six wells in Luzhou. Quartz was the dominant mineral type, (16.9–87.21%, average 51.33%), followed by illite, calcite, dolomite, and pyrite. Our study revealed that (1) quartz content showed a moderate positive correlation with the total organic carbon (TOC) content, indicating that the quartz found in the shale is mostly of biological origin; and (2) the sum content of siliceous minerals and carbonaceous minerals was moderately positively correlated with the brittleness index (BRIT) in well SS1H2-7 and in the well group of RS8 and RS5, indicating that the siliceous minerals and carbonaceous minerals had an active effect on reservoir compressibility. Finally, according to the mineralogical features of each sublayer, we identified four types of reservoirs to determine their scope for exploration.

In order to find out the enrichment mechanism and forming type of deep shale gas, taking the Longmaxi Formation shale in the Desheng–Yunjin Syncline area of Sichuan Basin as an example, we determined the mineralogy, organic geochemistry, physical property analysis, gas and water content, and the influence of three factors, namely sedimentation, structural conditions, and hydrogeological conditions, on the enrichment of shale gas. The results show that Longmaxi Formation shale in Desheng–Yunjin Syncline area is a good hydrocarbon source rock that is in the over-mature stage and has the characteristics of high porosity, low permeability, and high-water saturation. The contents of clay and quartz are high, and the brittleness index is quite different. According to the mineral composition, nine types of lithofacies can be found. The development characteristics of Longmaxi Formation shale and the sealing property of the roof have no obvious influence on the enrichment of shale gas, but the tectonic activities and hydrodynamic conditions have obvious influence on the enrichment of shale gas. The main control factors for shale gas enrichment in different regions are different. According to the main control factors, the gas accumulation in the study area can be divided into three types: fault-controlled gas, anticline-controlled gas, and hydrodynamic-controlled gas. The fault-controlled gas type is distributed in the north of the Desheng syncline and the north of the Yunjin syncline, the anticline-controlled gas type is distributed in the south of the Desheng syncline and the south of the Yunjin syncline, and the hydrodynamic-controlled gas type is distributed in the middle of the Baozang syncline. This result is of great significance for deep shale gas exploration.

ABSTRACT: Hydraulic fracturing induced seismicity has been frequently reported in the southern Sichuan Basin of southwest China, but the relationship between induced earthquakes and frictional stability-permeability of natural fractures in this area has not been well understood yet. Hydraulic fracturing process may change the friction-permeability relationship of fracture surface and thus has an important influence on the stability of fractures/faults. In this paper, we employed the friction -permeability coupling experiments on natural shale fractures, recovered from the 3700 m depth reservoir in the southern Sichuan Basin, to measure the friction coefficient and permeability change and clarify the friction-permeability-stability relationship of natural shale fractures. We carried out constant velocity shearing experiments, velocity step shearing experiments, constant velocity shear-flow experiments, and velocity step shear-flow experiments. In the experiments, the confining pressure is kept at 3 MPa, the constant velocity is set at μm/s and the velocity step are set at 10-1-10-1-10 μm/s. Results show that the fractures are frictional strong with the friction coefficient is 0.6 ~ 0.75. The friction coefficient of the same fracture in the pure shearing experiments is lower than that in the shear-flow experiments. The results of the velocity step experiments show that (a − b) >0 under all conditions, which is manifested as velocity strengthening, indicating aseismic slips possibly caused by the rich phyllosilicate content in the shale samples. We believe that the stability of the fracture is affected by the combination of factors such as the mineral composition, roughness of the fracture surface and the pressure of the injected fluid. 1. INTRODUCTION Seismic activity induced by fluid injection has attracted more and more attention, especially in hydraulic fracturing of shale gas exploitation (Igonin et al., 2021; Schultz et al., 2020; Grigoli et al., 2017; Elsworth et al., 2016; Bao & Eaton, 2016; Ellsworth, 2013). Sichuan Basin is rich in shale gas resources. Since 2009, the initial exploration has gradually entered the stage of large-scale exploitation while hydraulic fracturing is the main stimulation technology (Zou et al., 2016; Liang et al., 2012). However, the complex geological conditions of the Sichuan Basin make natural faults and fractures extensively developed in this area (Charvet, 2013). The characteristics of natural fractures in the deeper Longmaxi shale and peripheral faults have an important impact on the safe and efficient exploitation of shale gas (Atkinson et al., 2020; Wang & Shen, 2020; Zhou et al., 2012). With the increasing concerns of induced seismicity in recent years, it is necessary to study the friction-stability-permeability of natural fracture surfaces (Fang et al., 2017; Zhang et al., 2017; Ikari et al., 2011; Crawford et al., 2008; Yasuhara et al., 2004) for Longmaxi shale (Kohli & Zoback, 2013).

The optimal design of hydraulic fracturing parameters is the key to commercial exploitation of unconventional reservoirs. Hydraulic fracturing test is one of the main methods for optimizing fracturing parameters. It is known that scale effect exists between laboratory experiments and field treatments of hydraulic fracturing. However, studies on how to eliminate the scale effect are rarely reported. In this work, we conduct sensitivity analysis on rock mechanical parameters and fracturing parameters at different scales by using the dimensionless analysis method. The initiation and propagation process of field hydraulic fracturing is reproduced through laboratory tests, and fracturing parameters are analyzed by using numerical simulation. Our results show that the fracture propagation in the laboratory is inconsistent with that in the field fracturing. The fracture initiation and propagation in the field can be reproduced in experiments by using samples with high modulus and low toughness as well as high-viscosity fracturing fluid. Microcracks are created before the breakdown pressure is reached, and hydraulic fractures extend perpendicular to the direction of the minimum principal stress. The Carter’s leak-off coefficient has little effect on breakdown pressure and propagation pressure, but the injection rate and the horizontal principal stress have significant effects on breakdown pressure. This study provides a theoretical basis and guidance for the design of fracturing parameters both in the laboratory and in the field.

The characteristics of lithofacies in the shale reservoir of the Wufeng and Longmaxi formations located in the Luzhou gas field were studied using the three-end-member method (mineralogical components). In addition, the microscopic characteristics and pore genesis of different lithofacies were studied and compared using TOC, FESEM, and digital core images. The results of the present study showed the development of five lithofacies: 1) quartz-rich argillaceous shale; 2) quartz/clay-mixed shale; 3) calcareous/clay-mixed shale; 4) calcium-rich argillaceous shale; and 5) clay-rich siliceous shale. The degree of pore development between lithofacies was highly heterogeneous. In addition, data indicated that total shale porosity increased with the increase in TOC. Thus, the pores were mainly related to organic matter. The enrichment of framboidal pyrite in quartz-rich shale (S1, S-3) and mixed shale facies (M-2, M-3) is essential during the formation of high-quality reservoirs. On the other hand, the content of organic matter in clayey shale was low, which does not favor the development of high-quality reservoirs. The differences in lithofacies pore structures are controlled by the sedimentary environment. The strong retention and reduction environment are the most favorable features for the formation of organic-rich siliceous shale lithofacies, which promote the preservation of organic matter and the development of reservoirs.

Neural machine translation (NMT) optimized by maximum likelihood estimation (MLE) usually lacks the guarantee of translation adequacy. To alleviate this problem, we propose an NMT approach that heightens the adequacy in machine translation by transferring the semantic knowledge from bilingual sentence alignment learning. Specifically, we first design a discriminator that learns to estimate sentence aligning score over translation candidates. The discriminator is constructed by gated self-attention based sentence encoders and trained with an N-pair loss for better capturing lexical evidences from bilingual sentence pairs. Then we propose an adversarial training framework as well as a sentence alignment-aware decoding method for NMT to transfer the discriminator’s learned semantic knowledge to NMT models. We conduct our experiments on Chinese → English, Uyghur → Chinese and English → German translation tasks. Experimental results show that our proposed methods outperform baseline NMT models on all these three translation tasks. Further analysis also indicates the characteristics of our approaches and details the semantic knowledge that transfered from the discriminator to the NMT model.

A positive-gate-bias stress (PBS)-induced hump phenomenon in staggered bottom-gate amorphous indium–gallium–zinc oxide (a-IGZO) thin-film transistors (TFTs) is observed, and the possible reason is discussed. It is found that the hump phenomenon is closely related to device layouts. In conventional metal-over-active (MOA) structure, the hump current is independent of the channel width. However, for the active-over-metal (AOM) structure, the hump current disappears when the width of active layer exceeds a critical value. These results indicate that the hump current caused by PBS comes from parasitic edge transistors rather than back-channel transistors. Furthermore, the possible origin of channel edge conduction mechanism is checked by atomic force microscopy (AFM) and X-ray photoelectron spectroscopy (XPS) techniques. From AFM image, we notice that there is a long gentle slope at edge of active layer, which will enhance the electrical filed at local area and accelerate the device performance degradation under PBS condition. XPS measurements before and after PBS on devices at channel edge confirm our assumption, that is, the oxygen vacancy ( $\text{O}_{\mathrm {V}}$ ) content increases from 29.33% to 36.89% after PBS. This implies that the carrier concentration in channel edge is enhanced resulting in lower turn-on voltage of parasitic edge transistors, which, in turn, causes hump phenomenon.

In atomically-thin two-dimensional (2D) semiconductors, the nonuniformity in current flow due to its edge states may alter and even dictate the charge transport properties of the entire device. However, the influence of the edge states on electrical transport in 2D materials has not been sufficiently explored to date. Here, we systematically quantify the edge state contribution to electrical transport in monolayer MoS2/WSe2 field-effect transistors, revealing that the charge transport at low temperature is dominated by the edge conduction with the nonlinear behavior. The metallic edge states are revealed by scanning probe microscopy, scanning Kelvin probe force microscopy and first-principle calculations. Further analyses demonstrate that the edge-state dominated nonlinear transport shows a universal power-law scaling relationship with both temperature and bias voltage, which can be well explained by the 1D Luttinger liquid theory. These findings demonstrate the Luttinger liquid behavior in 2D materials and offer important insights into designing 2D electronics., Though metallic edge states in monolayer transition metal dichalcogenide have been observed before, how these states contribute to charge transport remains unclear. Here, the authors quantify the edge state contribution to electrical transport in monolayer MoS2/WSe2 field-effect transistors, revealing a dominated non-linear edge transport at low temperature, indicating possible Luttinger liquid behavior.

In this study, shale cores from 20 wells in the S1l11-1 sublayer of Longmaxi Formation buried in shallow shale (3500 m) in the southern Sichuan Basin, China were collected to compare their pore structures and gas-bearing properties using multiple experiments. Results showed that the deep layer has relatively lower brittle mineral content, which is disadvantageous in terms of the higher requirements it imposes on hydraulic fracturing. Results also showed that the most important factor controlling the differential enrichment of S1l11-1 shale gas in southern Sichuan Basin is porosity. Moreover, the porosity composition of shallow shale and deep shale has significant differences: the porosity of shallow shale is dominated by organic pores, while for deep shale, both organic and inorganic pores are important. The inorganic pores provide significant storage space for free gas in deep shale; their contribution warrants more attention. We also found that the difference in organic porosity of the shallow and deep shale samples resulted from large differences in pore development ability, while the highest inorganic porosity was concentrated near the optimal mineral composition when the content of quartz plus feldspar plus pyrite was about 70%. This study revealed the primary factor controlling the difference in gas content between shallow and deep shale and detailed the characteristics of microscopic pore structure, providing a basis for the exploration and development of deep shale gas in the Wufeng-Longmaxi Formation in the southern Sichuan Basin.

We present a new methodology to estimate the lifetime of thin-film-transistor (TFT)-based scan driver circuit in display application with the dynamic time evolutionary aging model. The proposed method not only enables more accurate device level estimation but also circuit level aging prediction given that the TFTs are stressed dynamically. To properly handle threshold voltage shift ( $\Delta \text{V}_{\text {th}}$ ) of each TFTs in the scan driver under dynamic voltage stress patterns like the alternative voltage level and frequency, a time discretization strategy considering the voltage bias stress feedback is developed to be adaptable with SPICE transient simulation simply. The results show that $\Delta \text{V}_{\text {th}}$ of each TFTs in scan driver circuit can be evolved in real time accordingly under dynamic stress. What’s more, an exponential relationship between the aging parameter $\beta $ and the aging time is for the first time obtained, which has much importance for predicting lifetime of scan driver and other TFT-based circuit.

In this paper, we demonstrate a tunable Schottky diode based on amorphous indium–gallium–zinc–oxide (a-IGZO) film with high performance for rectifying circuit and bipolar one diode-one resistor (1D1R) crossbar applications. By modulating the oxygen content in the a-IGZO film, the device shows adjustable behavior. Under the high oxygen content condition, the device displays outstanding diode features with a large rectification ratio up to 108 at ±1 V and a high forward current density of 100 A/cm2 at ±1 V when used Cu as the top electrode. The thermionic emission model is proposed to explain the Schottky diode behavior. And by utilizing the superior rectifying performance, we successfully demonstrate the dynamic rectifying application. In addition, we combine the optimized Schottky diode device with the bipolar-type resistive random access memory (RRAM) device to investigate the Schottky diode device in 1D1R crossbar application. As a result of its tunable nature and low processing temperature, the a-IGZO Schottky diode is a promising element for rectifying circuit and bipolar 1D1R crossbar applications.

For the first time, in this letter, the anomalous positive bias stress instability of the back gated monolayer polycrystal molybdenum disulfide (MoS2) field-effect transistors with high-hydrogen-concentration SiO2 gate dielectrics is reported. It is found that the threshold voltage shifts exhibit a pronounced turnaround behavior from a positive shift to a negative shift when enlarging the stress voltage and stress time. We relate this anomalous positive bias instability to two physical processes, charge trapping and hydrogen release and migration.

In this letter, we demonstrate the coexistence of Schottky diode performance and resistance switching (RS) behavior based on TiN/amorphous indium-gallium-zinc oxide/Cu device. Depending on the reverse voltage applied to the Cu electrode, the device itself can be obviously switched from the diode mode to the RS mode. The device shows outstanding diode features with a large rectification ratio up to $ {3}\times {10}^{ {6}}$ at ±1 V and a high forward current density of 100 A/cm2 at +1 V. Furthermore, repeatable and stable switching behavior with low voltage has also been observed by increasing the reverse forming voltage. The physical model of Schottky barrier and conductive filament is proposed to explain the coexistence phenomenon. These achievements increase the application diversity of the metal–oxide film and provide a great potential application in active-matrix flat-panel displays.

To optimise the performance of photo-field-effect transistor, a back-gate hybrid structure was developed. In this hybrid phototransistor, lead sulphide (PbS) thin film was prepared using physical vapour deposition as a photosensitive layer. Discontinuous and uniform PbS film was obtained by controlling the deposition rate and time of PbS powders. Amorphous indium gallium zinc oxide (IGZO) with high mobility was used as an active layer. In this work, The hybrid structure phototransistor shows an excellent performance: device mobility (μ) reach 8.7 cm2V−1s−1, and responsivity achieve 2.7 × 104 A/W in visible spectrum and 5.7 A/W in near-infrared spectrum, respectively. Furthermore, the transistor exhibits detectivity up to 2.79 × 1013 cmHz1/2W−1. The device also exhibits characteristics of the ideal diode: the saturation current of the diode is as small as 0.422 nA, and the responsivity of diode is ∼0.74 A/W. Simplified manufacturing processes effectively reduce the cost of fabricated device and provide better device stability.

Ascertaining the characteristics of fracture system at different scales integratedly is very important for performing efficient exploration and development activities of specific shale gas reservoirs. In this paper, an area around 250 square kilometers in Changning Block of Sichuan Basin is taken as an example, which belongs to Chinese Shale Gas Development Demonstration Plot. Seismic structural interpretation was performed detailedly based on original seismic amplitude cube and derived edge-detection cubes, and then the technologies of finite element horizon flattening, orthogonal decomposition principal component analysis, seismic discontinuity patch auto-extraction and paleo-stress field inversion were applied, together with the existing regional geological understanding and fracture information in wells, to figure out the staging and grouping of fracture system at seismic scale (i.e., at large and middle scales), at the same time to clarify the regional tectonic evolution and its genetic relationship with fractures at different scales such as the ones revealed by seismic data and cores or image logs. The following conclusions were reached. (a) The tectonic movements affecting the development of fracture system in study interval mainly happened during Yanshanian-Himalayan periods, i.e., 3 compressional tectonic episodes which were nearly in S-N direction in Late Yanshanian period, in NNE-SSW direction in Early Himalayan period, and in NWW-SEE direction in Middle Himalayan period respectively. (b) The Late Yanshanian tectonic event primarily formed long-axis anticlines and synclines, thrust faults and fault-related fractures, all of which were nearly in E-W trending, and fold-related fractures in different directions. (c) The Early Himalayan tectonic event mainly formed genetically related conjugate fracture sets including strike-slip faults and shear fractures both in NNW and NE directions, and transverse extensional fractures. (d) The Middle Himalayan tectonic event chiefly formed thrust faults, and related fractures and folds in NNE~NE direction, and transverse extensional fractures. (e) Furtherly our work demonstrated that such kind of fracture system analysis was of great significance in building discrete fracture network, providing precautionary advice for drilling engineering, and optimizing completion program and field development plan, etc. Hence, integrated fracture system analysis at full scales to reach more meaningful and robust conclusions is essential work for unconventional resources evaluation and characterization.

Most state-of-the-art QE systems built upon neural networks have achieved promising performances on benchmark datasets. However, the performance of these methods can be easily influenced by the inherent features of the model input, such as the length of input sequence or the number of unseen tokens. In this paper, we introduce a causal inference based method to eliminate the negative impact caused by the characters of the input for a QE system. Specifically, we propose an iterative denoising framework for multiple confounding features. The confounder elimination operation at each iteration step is implemented by a Half-Sibling Regression based method. We conduct our experiments on the official datasets and submissions from WMT 2020 Quality Estimation Shared Task of Sentence-Level Direct Assessment. Experimental results show that the denoised QE results gain better Pearson’s correlation scores with human assessments compared to the original submissions.

Neural machine translation (NMT) usually employs beam search to expand the searching space and obtain more translation candidates. However, the increase of the beam size often suffers from plenty of short translations, resulting in dramatical decrease in translation quality. In this paper, we handle the length bias problem through a perspective of causal inference. Specifically, we regard the model generated translation score S as a degraded true translation quality affected by some noise, and one of the confounders is the translation length. We apply a Half-Sibling Regression method to remove the length effect on S, and then we can obtain a debiased translation score without length information. The proposed method is model agnostic and unsupervised, which is adaptive to any NMT model and test dataset. We conduct the experiments on three translation tasks with different scales of datasets. Experimental results and further analyses show that our approaches gain comparable performance with the empirical baseline methods.

In order to solve the problem of the identification accuracy and eeg signal stability of BCI equipment, firstly, the identification accuracy, signal-to-noise ratio and stability of BCI to EEG signals are improved through the study of the pre-processing, feature extraction of motion imagination EEG signals, so as to obtain more characteristic information.Secondly, wavelet thresholding is used as the pre-processing method of EEG signal.Finally, the feature extraction of eeg signals from motion imagination.The experimental results show that the expected EEG signal can be denoised and feature extracted.

In this paper, the EEG signal processing and wheelchair control research.Firstly, the EEG signal is preprocessed and the feature of motion image EEG signal is extracted.Boosting and Boosting two ensemble learning algorithms, which include AdaBoost and Gradient Boosting, both Boosting algorithms, were Boosting, Boosting and Boosting.Based on Three algorithms, Bagging, AdaBoost and Gradient Boosting, three classifiers are constructed, and the feature vectors are classified and identified.Finally, the recognized signals are converted into external instructions to realize wheelchair motion control. The experimental results verify the effectiveness of the proposed EEG signal feature extraction algorithm and classification recognition algorithm.

Even as an important lexical information for Latin languages, word case is often ignored in machine translation. According to observations, the translation performance drops significantly when we introduce case-sensitive evaluation metrics. In this paper, we introduce two types of case-sensitive neural machine translation (NMT) approaches to alleviate the above problems: i) adding case tokens into the decoding sequence, and ii) adopting case prediction to the conventional NMT. Our proposed approaches incorporate case information to the NMT decoder by jointly learning target word generation and word case prediction. We compare our approaches with multiple kinds of baselines including NMT with naive case-restoration methods and analyze the impacts of various setups on our approaches. Experimental results on three typical translation tasks (Zh-En, En-Fr, En-De) show that our proposed methods lead to the improvements up to 2.5, 1.0 and 0.5 in case-sensitive BLEU scores respectively. Further analyses also illustrate the inherent reasons why our approaches lead to different improvements on different translation tasks.

SummaryCurrently, there is large-scale shale gas exploration and development in the Sichuan Basin, western China. Caused by high tectonic stress and presence of fracture systems at various scales in the lower Silurian Longmaxi reservoir formation, hydraulic fracturing in shale gas reservoirs in the Sichuan Basin has encountered many difficulties, such as placing sufficient proppant, poor-production performance for some wells, and ambiguity as to the factors controlling the production of reservoirs. It has been recognized that lack of geomechanical understanding of the shale gas reservoirs is a major obstacle to effectively addressing these difficulties.A 3D full-field geomechanics model was constructed for the Changning shale gas reservoir in the Sichuan Basin through integrating seismic, geological structure, log, and core data by following a newly formulated work flow. The 3D geomechanical model includes 3D anisotropic mechanical properties, 3D pore pressure, and the 3D in-situ stress field. Through leveraging measurements from an advanced sonic tool and core data, the anisotropy of the formation was captured at wellbores and propagated to 3D space guided by prestack seismic inversion data. The 3D pore-pressure prediction was conducted with seismic data, and calibrated against pressure measurements, mud-logging data, and flowback data. A discrete-fracture-network (DFN) model, which represents multiscale natural-fracture systems, was integrated into the 3D geomechanical model during stress modeling to reflect the disturbance on the in-situ stress field by the presence of the natural-fracture systems.The 3D pore-pressure model was used to calculate more-reliable estimates of gas in place in the shale gas reservoir, and the geomechanical model was used to reveal the root cause of difficulties of proppant placement in this tectonically active and unevenly fractured shale gas reservoir.The paper presents the highlights and innovations in constructing the 3D geomechanical model for the shale gas reservoir, and explains how the 3D geomechanical model is used to address technical challenges encountered during drilling and completion. Also, it demonstrates that a reliable 3D geomechanical model, with proper characterization of anisotropy, pore pressure, and natural fractures, provides a critical opportunity to improve the development in this shale gas reservoir.

Hashtags of microblogs can provide valuable information for many natural language processing tasks. How to recommend reliable hashtags automatically has attracted considerable attention. However, existing studies assumed that all the training corpus crawled from social networks are labelled correctly, while large sample statistics on real social media shows that there are 8.9% of microblogs with hashtags having wrong labels. The notable influence of noisy data to the classifier is ignored before. Meanwhile, recency also plays an important role in microblog hashtag, but the information is not used in the existing studies. Some temporal hashtags such as World Cup will ignite at a particular time, but at other times, the number of people talking about it will sharply decrease. To address the twofold shortcomings above, the authors propose an long short-term memory-based model, which uses temporal enhanced selective sentence-level attention to reduce the influence of wrong labelled microblogs to the classifier. Experimental results using a dataset of 1.7 million microblogs collected from SINA Weibo microblogs demonstrated that the proposed method could achieve significantly better performance than the state-of-the-art methods.

An analytical Seebeck coefficient model for disordered organic semiconductors based on hopping transport and percolation theory is proposed here. This model demonstrates the relationships between Seebeck coefficient and temperature, carrier concentration as well as disorder degree of materials. As compared with experimental data, the simulated results show a convincing coincidence with experimental results. Moreover, the effect from doping is addressed. The calculation results show that the Seebeck coefficient will decrease with increasing doping ratio, after passing a minimum value then a sharp increase of Seebeck coefficient appears.

Recently, topic modeling has been widely used to discover the abstract topics in the multimedia field. Most of the existing topic models are based on the assumption of three-layer hierarchical Bayesian structure, i.e. each document is modeled as a probability distribution over topics, and each topic is a probability distribution over words. However, the assumption is not optimal. Intuitively, it’s more reasonable to assume that each topic is a probability distribution over concepts, and then each concept is a probability distribution over words, i.e. adding a latent concept layer between topic layer and word layer in traditional three-layer assumption. In this paper, we verify the proposed assumption by incorporating the new assumption in two representative topic models, and obtain two novel topic models. Extensive experiments were conducted among the proposed models and corresponding baselines, and the results show that the proposed models significantly outperform the baselines in terms of case study and perplexity, which means the new assumption is more reasonable than traditional one.

The Lungmachi Formation is widely distributed in Guizhou, Chongqing and the adjacent area. It is important for the study of Silurian biostratigraphy and shale-gas investigation. Based on those biostratigraphically well-studied sections from Guiyang to Huayingshan, we reveal the stage-progressive distribution pattern of the Lungmachi black shales. The distribution of the Lungmachi black shales in the studying area can be subdivided into four geographic belts from the south to the north, reflecting the joint effect of regional and global environmental changes. The graptolite depth zonation model was adopted herein to infer the water depth of major graptolite assemblages from the black shales. The changes in the water depth indicate two major stages. The first stage is named the transgressive distribution stage which ranged from the Persculptograptus persculptus Biozone (LM1, upper Hirnantian) to the Coronograptus cyphus Biozone (LM5, upper Rhuddanian), an interval mostly controlled by global sea-level rise. The second stage, ranging from the Demirastrites triangulatus Biozone (LM6, lower Aeronian) to the Spirograptus guerichi Biozone (LM9, lower Telychian), is named the regressive shrinking stage, during which the black shales were gradually replaced by mixed-facies or carbonate sediments from the south to the north, representing the effects of the persistent uplifting of the Central Guizhou Oldland.

Ning201 Longmaxi shale gas play is located on the southwest edge of Sichuan Basin. Block Ning201 surface is very mountainous and elevation range of drilling pads is between 400 meters and 1300 meters. The target layer of shale gas exploration and development is mainly the Wufeng Formation of the Ordovician system and the organic-rich shale interval of the Lower Longxixi Formation of the Silurian system. The total thickness is between 30m and 50m, and the overall buried depth is 2300~3200m. Field development of Ning201 shale gas play was geared up in 2014. During early appraisal phase before 2014, average testing results of wells were 10~15×104 m3/d. 7 vertical appraisal wells and 50 horizontal wells have been drilled and 38 horizontal wells among them have been put into production to evaluate optimized development strategy until middle of 2017. Changning area accomplished the construction of 15×108 m3/a production capacity. During production appraisal phase, different well spacing pads were configured, and some cross-well interference testing were performed to determine optimal spacing in Ning201 block. Only qualitatively understanding can be achieved with spontaneous field efforts and quantitative evaluation of long-term impacts on production with varying initial rate control and well spacing is still missing. In this paper, an integrated modeling workflow was introduced to accurately simulate the production behavior of a representative shale gas well and to control production through a calibration model to optimize well spacing and maximize economic returns. The workflow begins with microseismic monitoring during shale gas well fracturing operations. Considering the fracturing pumping procedure, fluid and proppant volume and subsurface features, including formation, natural fractures and in-situ stresses, the heterogeneity of the hydraulic fracturing system propagation ensures that the simulated complex hydraulic fracturing network is in line with reality. After establishing a detailed hydraulic fracturing model, production performance is matched to history to better calibrate the conductivity of hydraulic fracture network in dynamic model as a relaible predicting tool.

Picture news has become more and more popular among online news in recent years. As the first impression to viewers, thumbnail plays a very important role in picture news. However, it is time consuming to manually select thumbnails for a huge amount of picture news. In this paper, we introduce a new task of automatic picture news thumbnail selection. Given a piece of picture news containing a set of images, this task is to select several appropriate images from the picture news as candidate thumbnails. To this end, we present a large publicly available image dataset for this task, called Picture News Collection(The Picture News Collection 0.1 version can be publicly available online at https://github.com/anonymity01/Picture-News-Collection.). The Picture News Collection contains more than 4 million images of 347,731 picture news from two famous news websites, Sina News and NetEase News. Selecting good enough thumbnails is complicated and needs to consider many aspects, such as attraction, hot topics, content integrity, etc. In order to select appropriate candidate thumbnails, we propose an attention-based thumbnail selection model, and the experimental results comparing with three image classification based baselines show that our proposed methods outperform the baselines. We introduce the automatic picture news thumbnail selection task and the dataset to encourage further studies of this challenge.

Neural Machine Translation (NMT) optimized by Maximum Likelihood Estimation (MLE) lacks the guarantee of translation adequacy. To alleviate this problem, we propose an NMT approach that heightens the adequacy in machine translation by transferring the semantic knowledge learned from bilingual sentence alignment. Specifically, we first design a discriminator that learns to estimate sentence aligning score over translation candidates, and then the learned semantic knowledge is transfered to the NMT model under an adversarial learning framework. We also propose a gated self-attention based encoder for sentence embedding. Furthermore, an N-pair training loss is introduced in our framework to aid the discriminator in better capturing lexical evidence in translation candidates. Experimental results show that our proposed method outperforms baseline NMT models on Chinese-to-English and English-to-German translation tasks. Further analysis also indicates the detailed semantic knowledge transfered from the discriminator to the NMT model.

In this paper, we transform tag recommendation into a word-based text generation problem and introduce a sequence-to-sequence model. The model inherits the advantages of LSTM-based encoder for sequential modeling and attention-based decoder with local positional encodings for learning relations globally. Experimental results on Zhihu datasets illustrate the proposed model outperforms other state-of-the-art text classification based methods.