A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Filters
Neural Text Generation: Past, Present and Beyond
[article]
2018
arXiv
pre-print
Preserved Fulltext
Besides, Self-BLEU [Zhu et al., 2018] , which calculates the BLEU score between the generated sentences, is used to monitor the severity of mode collapse. ...
BLEU [Papineni et al., 2002] and NLL test [Zhu et al., 2018] are used to evaluate the similarity between documents or generator's capacity to fit real data. ...
arXiv:1803.07133v1
fatcat:brfobli56nf6pn3kzhno5dwnea
Unified Multimodal Punctuation Restoration Framework for Mixed-Modality Corpus
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
For example, Yaoming and Liwei have equal contribution "I don't want anymore kids" means far away from "I don't want anymore, kids", suggesting the importance of a comma. ...
arXiv:2202.00468v1
fatcat:abruumq7tjfzlfdku3icr7mh7q
Texygen: A Benchmarking Platform for Text Generation Models
[article]
2018
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
We introduce Texygen, a benchmarking platform to support research on open-domain text generation models. Texygen has not only implemented a majority of text generation models, but also covered a set of metrics that evaluate the diversity, the quality and the consistency of the generated texts. The Texygen platform could help standardize the research on text generation and facilitate the sharing of fine-tuned open-source implementations among researchers for their work. As a consequence, this
arXiv:1802.01886v1
fatcat:buq2g3knjbbirigzdgflgejatm
more »
... ld help in improving the reproductivity and reliability of future research work in text generation.
GIKT: A Graph-based Interaction Model for Knowledge Tracing
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
With the rapid development in online education, knowledge tracing (KT) has become a fundamental problem which traces students' knowledge status and predicts their performance on new questions. Questions are often numerous in online education systems, and are always associated with much fewer skills. However, the previous literature fails to involve question information together with high-order question-skill correlations, which is mostly limited by data sparsity and multi-skill problems. From
arXiv:2009.05991v1
fatcat:b3afj23hj5aedmvyesyx77o7ve
more »
... e model perspective, previous models can hardly capture the long-term dependency of student exercise history, and cannot model the interactions between student-questions, and student-skills in a consistent way. In this paper, we propose a Graph-based Interaction model for Knowledge Tracing (GIKT) to tackle the above probems. More specifically, GIKT utilizes graph convolutional network (GCN) to substantially incorporate question-skill correlations via embedding propagation. Besides, considering that relevant questions are usually scattered throughout the exercise history, and that question and skill are just different instantiations of knowledge, GIKT generalizes the degree of students' master of the question to the interactions between the student's current state, the student's history related exercises, the target question, and related skills. Experiments on three datasets demonstrate that GIKT achieves the new state-of-the-art performance, with at least 1% absolute AUC improvement.
Counter-Interference Adapter for Multilingual Machine Translation
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Other Versions
., 2019; Zhu et al., 2020) .
Adapter Network for Machine Translation. Our design derives from the residual adapters of the domain adaptation task. ...
en2ar en2fr en2de en2zh en2ru ar2en fr2en de2en zh2en ru2en
Direction
Average Cosine Similarity
model
w/ emb adapter
w/o emb adapter
Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu ...
arXiv:2104.08154v2
fatcat:lysh3fhjhfduvm4zxyzmyql5oq
The Volctrans Machine Translation System for WMT20
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Note that this fulltext copy is not of the "primary" version of this work. The version it corresponds to is:
2020 pre-print arXiv:2010.14806v1 fatcat:ztbvg4edhjhkxmusdd4b7dr7yi
This paper describes our VolcTrans system on WMT20 shared news translation task. We participated in 8 translation directions. Our basic systems are based on Transformer, with several variants (wider or deeper Transformers, dynamic convolutions). The final system includes text pre-process, data selection, synthetic data generation, advanced model ensemble, and multilingual pre-training.
arXiv:2010.14806v2
fatcat:ww4w67d6u5bypb46ym2bijmlpi
Progress in Electrolyte-Free Fuel Cells
2016
Frontiers in Energy Research
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2018; you can also visit the original URL.
The file type is application/pdf.
The different working principle of EFFC has been presented by Zhu et al. (2011c Zhu et al. ( , 2012 . ...
The bilayer EFFC device was discovered before Zhu et al. (2011e) . ...
Copyright © 2016 Lu, Zhu, Cai, Kim, Wang, Wang, Zhang and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). ...
doi:10.3389/fenrg.2016.00017
fatcat:gnnhivn3mzfbdeiy256ggfysmm
An Analog of the Adjugate Matrix for the Outer InverseAT,S(2)
2012
Mathematical Problems in Engineering
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2019; you can also visit the original URL.
The file type is application/pdf.
We investigate the determinantal representation by exploiting the limiting expression for the generalized inverseAT,S(2). We show the equivalent relationship between the existence and limiting expression ofAT,S(2)and some limiting processes of matrices and deduce the new determinantal representations ofAT,S(2), based on some analog of the classical adjoint matrix. Using the analog of the classical adjoint matrix, we present Cramer rules for the restricted matrix equationAXB=D, R(X)⊂T, N(X)⊃S∼.
doi:10.1155/2012/591256
fatcat:i7fbbefzy5esvf5vlzlow7dex4
Anisotropic vanadium dioxide sculptured thin films with superior thermochromic properties
2013
Scientific Reports
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2019; you can also visit the original URL.
The file type is application/pdf.
VO 2 (M) STF through reduction of V 2 O 5 STF was prepared. The results illustrate that V 2 O 5 STF can be successfully obtained by oblique angle thermal evaporation technique. After annealing at 5506C/3 min, the V 2 O 5 STF deposited at 856 can be easily transformed into VO 2 STF with slanted columnar structure and superior thermochromic properties. After deposition SiO 2 antireflective layer, T lum of VO 2 STF is enhanced 26% and DT sol increases 60% compared with that of normal VO 2 thin
doi:10.1038/srep02756
pmid:24067743
pmcid:PMC3783039
fatcat:ibx7joquznbdxhqoxpmpgahmcm
more »
... s. Due to the anisotropic microstructure of VO 2 STF, angular selectivity transmission of VO 2 STF is observed and the solar modulation ability is further improved from 7.2% to 8.7% when light is along columnar direction. Moreover, the phase transition temperature of VO 2 STF can be depressed into 54.56C without doping. Considering the oblique incidence of sunlight on windows, VO 2 STF is more beneficial for practical application as smart windows compared with normal homogenous VO 2 thin films. M onoclinic/rutile (M/R)-phase VO 2 , due to its reversible metal-insulator (MIT) phase transitions, has attracted great attention in many fields, such as resistive switching elements 1 , smart windows 2 , light modulators 3 and optical storage devices 4 , etc. With the energy supply tension and the rise of energysaving technology, VO 2 -based smart window becomes the focus due to its optical modulation ability in infrared spectra region. However, a low luminous transmittance (T lum ) and solar modulating ability (DT sol ) are the two major drawbacks for the practical application of VO 2 -base smart windows 5 . The T lum and T sol values were obtained from T lum,sol (t) 5 #Q lum,sol (l)T(l,t)dl/#Q lum,sol (l)dl, where Q lum is the spectral sensitivity of the light-adapted eye and Q sol is the solar irradiance spectrum for an air mass 1.5 (corresponding to the sun standing 37u above the horizon). DT sol is obtained from DT sol 5 T sol,l 2 T sol,h , where l and h denote low-and hightemperature, respectively. Different strategies are investigated to improve T lum and DT sol , including Mg-or F-doping 6,7 , multilayer-stack design 8,9 , composite film construction 10,11 , nanoporous incorporation 12 . Mg-or F-doping causes both the depression of transition temperature and a blue shift in the absorption edge of VO 2 films (from around 445 to 415 nm 6 ). A blue shift of absorption edge results in a significant increase of luminous transmittance with the expense of infrared modulating ability 7 . The competitive relation between luminous transmittance and solar modulation ability has also been observed in VO 2 based multilayer films 8 . A VO 2 -SiO 2 composite films show a high luminous transmittance but weak infrared modulating ability 10 . Optical calculations suggest that VO 2 nanoparticles distributed in a transparent dielectric matrix can show higher T lum and DT sol 13 . Gao et al. 11 have proved that VO 2based composite thin films show superior luminous transmittance and solar modulation ability. However, compared with other dielectric matrix, porosity shows much more advantages, such as low refractive index, good compatibility and flexible adjustability. Kang et al. 12 have proved that nanoporous VO 2 thin films prepared by spinning method can enhance the luminous transmittance and solar modulation ability. However, investigation on nanoporous VO 2 thin films prepared by physical vapor deposition is seldom. Oblique angle deposition (OAD) technique is a common technique to adjust the porosity and microstructure of thin films, which is named sculptured thin films (STF) 14 . Sculptured thin films with controllable porosity and microstructure can show gradual optical constants and special physical properties, which can be used in many fields, such as rugate filters 15 , sensors 16 , phase retarder 17 , etc. The adjustable optical constants of STF are beneficial for the optical design of multilayer thin films and the dielectric layer is more optional. The anisotropic OPEN
Climate-related risks in the construction of Xiongan New Area, China
2020
Theoretical and Applied Climatology
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
The self-purification capacity of the atmosphere is calculated using the method of Zhu et al. (2018) . ...
been shown that Xiongan New Area is experiencing a warming and drying trend (Ma and Ren 2007) , evidenced by significantly increased temperatures and decreased precipitation in the Hai River Basin (Zhu ...
doi:10.1007/s00704-020-03277-2
fatcat:kk4a3op7bjdino7aceqdukwbna
Dual monoclonal antibody-based sandwich ELISA for detection of in vitro packaged Ebola virus
2018
Diagnostic Pathology
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2019; you can also visit the original URL.
The file type is application/pdf.
Rapid transmission and high mortality of Ebola virus disease (EVD) highlight a urgent need of large scale, convenient and effective measure for Ebola virus screening. Application of monoclonal antibodies (mAbs) are crucial for establishment of an enzyme-linked immunosorbent assay (ELISA) with high sensitivity and specificity. Methods: The traditional cell fusion technique was used to generate a panel of hybridomas. Two mAbs were characterized by SDS-PAGE, Western blot, Indirect
doi:10.1186/s13000-018-0773-1
fatcat:oiuf7voa4zcz3puo455bz7vabu
more »
... e assay (IFA). A sandwich ELISA was established using the two mAbs. The detection capability of the ELISA was evaluated. Results: In the current study, we produced two murine-derived mAbs (designated as 6E3 and 3F21) towards Zaire Ebola virus glycoprotein (GP), the major viral transmembrane spike protein associated with viral attachment. It was shown that 6E3 and 3F21 recognized GP1 and GP2 subunits of the GP respectively. Furthermore, 6E3 and 3F21 bound to corresponding epitopes on GP without reciprocal topographical interpretation. Subsequently, a sandwich ELISA based on the two mAbs were established and evaluated. The detection limit was 3.6 ng/ml, with a linear range of 3.6-100 ng/ml. More importantly, Ebola virus like particles (eVLPs) were able to be detected by this established virus detection measure. Conclusions: We produced and characterized two murine-derived mAbs (designated as 6E3 and 3F21) towards Zaire Ebola virus glycoprotein (GP), and established a sandwich ELISA based on the mAbs. It was suggested that the sandwich ELISA provided an alternative method for specific and sensitive detection of Ebola virus in the field setting.
Development and Experimental Analysis of a Fuzzy Grey Control System on Rapeseed Cleaning Loss
2020
Electronics
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
One of the most important means of improving the mechanization of rapeseed harvests and increasing farmers' income is to reduce the cleaning loss of rapeseed. In this study, a fuzzy grey control system was developed using an assembled cleaning loss sensor. Based on experimental data, the relationship between the cleaning loss and the opening of the louver sieve in the cleaning device was obtained. The fuzzy control scheme was established by combining grey prediction and the fuzzy control
doi:10.3390/electronics9111764
fatcat:rrlngvixujb23jieoggnrq5tk4
more »
... le. Secondly, a microcontroller unit (MCU) was used as the controller, and the opening of the louver sieve was automatically regulated by detecting the signal of the cleaning loss. Finally, the performance and robustness of the control system was evaluated in field tests. Different experiments were conducted under different speed conditions to reflect the variable throughput. Results showed that using the grey prediction control system can realize the adjustment of the louver sieve opening in real time. The cleaning loss could be maintained within the ideal setpoint interval, compared with the operation with the control system switched off. These findings indicate that the application of the grey fuzzy control system reduces cleaning loss, and the nonlinear, time-variable and time delay problems in cleaning devices can be solved effectively.
Field Investigation of the Static Friction Characteristics of High-Yielding Rice during Harvest
2022
Agriculture
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Following the popularization of high-yielding rice in China, fast and efficient mechanized harvesting proved challenging. In addition, the physical characteristics of rice grains and stems are substantially affected during harvest by the field environment and harvest time. However, the combine harvester driver is focused on maximizing the outputs and does not consider the adverse effects of these factors during the rice harvest. Methods: We investigated the effects of the harvest time, spatial
doi:10.3390/agriculture12030327
fatcat:zafmzpossbbvrmmso4cjrimpge
more »
... osition, and temperature on the static friction coefficient of rice grains and stems of high-yielding rice using a field experiment. Results: The result difference in the static friction coefficient between the parallel and perpendicular placements of the rice stem on the steel plate was 9%, indicating that the contact configuration had a significant impact. The region, harvest time, and temperature significantly affected the static friction characteristics of the rice grains and stems. The most significant differences were observed in the X-direction. Conclusions: The optimum harvest time was 10:11 a.m.–3:30 p.m. and the optimum temperature was above 16.5 °C. A quantitative analysis of the effects of the harvest time and temperature on the static friction characteristics of rice provides reliable data for machine design optimization and standardization of harvests operations.
The Volctrans GLAT System: Non-autoregressive Translation Meets WMT21
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Other Versions
This paper describes the Volctrans' submission to the WMT21 news translation shared task for German->English translation. We build a parallel (i.e., non-autoregressive) translation system using the Glancing Transformer, which enables fast and accurate parallel decoding in contrast to the currently prevailing autoregressive models. To the best of our knowledge, this is the first parallel translation system that can be scaled to such a practical scenario like WMT competition. More importantly,
arXiv:2109.11247v2
fatcat:b5luur4ixvdejduxwmsznje4p4
more »
... parallel translation system achieves the best BLEU score (35.0) on German->English translation task, outperforming all strong autoregressive counterparts.
Signal Instructed Coordination in Cooperative Multi-agent Reinforcement Learning
[article]
2019
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Other Versions
In many real-world problems, a team of agents need to collaborate to maximize the common reward. Although existing works formulate this problem into a centralized learning with decentralized execution framework, which avoids the non-stationary problem in training, their decentralized execution paradigm limits the agents' capability to coordinate. Inspired by the concept of correlated equilibrium, we propose to introduce a coordination signal to address this limitation, and theoretically show
arXiv:1909.04224v2
fatcat:eifkmtj2sfgspp7gesqot2onay
more »
... t following mild conditions, decentralized agents with the coordination signal can coordinate their individual policies as manipulated by a centralized controller. The idea of introducing coordination signal is to encapsulate coordinated strategies into the signals, and use the signals to instruct the collaboration in decentralized execution. To encourage agents to learn to exploit the coordination signal, we propose Signal Instructed Coordination (SIC), a novel coordination module that can be integrated with most existing MARL frameworks. SIC casts a common signal sampled from a pre-defined distribution to all agents, and introduces an information-theoretic regularization to facilitate the consistency between the observed signal and agents' policies. Our experiments show that SIC consistently improves performance over well-recognized MARL models in both matrix games and a predator-prey game with high-dimensional strategy space.
« Previous
Showing results 1 — 15 out of 167 results