arXiv subjects

Hui Wan

Publications and source records attributed to Hui Wan.

At least 37 records · Page 2Linked to original sources

arXiv2022

Learning as Conversation: Dialogue Systems Reinforced for Information Acquisition

We propose novel AI-empowered chat bots for learning as conversation where a user does not read a passage but gains information and knowledge through conversation with a teacher bot. Our information-acquisition-oriented dialogue system employs a novel adaptation of reinforced self-play so that the system can be transferred to various domains without in-domain dialogue data, and can carry out conversations both informative and attentive to users. Our extensive subjective and objective evaluations on three large public data corpora demonstrate the effectiveness of our system to deliver knowledge-intensive and attentive conversations and help end users substantially gain knowledge without reading passages. Our code and datasets are publicly available for follow-up research.

cs.CL↗

arXiv2022

In-depth characterization and analysis of simple shear flows over regularly arranged micro pillars, II. Effect of pillar arrangement

Through high-fidelity numerical simulation, the effect of the arrangement of micropillars on the flow characteristics and momentum transport has been extensively investigated. The surface friction due to the complex flow characteristics and momentum transport mechanism has also been studied in depth. The micropillars are arranged in a quadrilateral, and different arrangements are acquired by changing the streamwise and spanwise distances between pillar rows. The results show that the streamwise and spanwise pillar distances have their own different influences. When the streamwise pillar distance is small, the micro eddies in the gaps between the streamwise neighboring pillars are significantly suppressed. The increase in the spanwise pillar distance enhances the momentum transport from the flow above pillar array to the flow in the spaces among micro pillars. When the spanwise pillar distance is small, the micro eddies in the gaps between the streamwise neighboring pillars connect with each other and form a tubular eddy between each pair of spanwise pillar rows. The tubular eddies significantly reduce the momentum transport from the upper flow to the lower flow. The increase in the streamwise pillar distance increase the momentum flux slightly. The surface friction can be decomposed into three components which are associated with two factors, the dilution effect of the number density of micro pillars and the multi-faceted effects of micro eddies. These two factors are determined by the streamwise and spanwise pillar distances. The dependence of the total friction and its components on the pillar distances has been thoroughly examined.

physics.flu-dyn↗

arXiv2022

In-depth characterization and analysis of simple shear flows over regularly arranged micro pillars, I. Effect of fluid inertia

Through high-fidelity numerical simulation, the simple shear flow over regularly arranged micro pillars has been investigated. The essential issues to be addressed include the characteristics of a simple shear flow over quadrilateral array of micro pillars, the effect of fluid inertia on the basic flow pattern, and the decomposition of the complex surface friction. The results show that the flow is characterized by a series of microscale recirculating eddies in the gaps between the streamwise neighboring pillars. The recirculation of the micro eddies and the oscillation of the overhead flow climbing over the pillar tips create a local flow advection. At smaller Reynolds number, the fluid inertia is weak and the flow patterns are symmetrical about the pillar center. When the Reynolds number is sufficiently large, the fluid inertia takes effect and breaks the symmetrical patterns. The overhead flow tilts downward, forming a spiral long-range advection between the fluid flow above pillar array and the flow in the spaces among micro pillars. The local advection and long-range advection constitute the transport mechanism in wall-normal direction. On micro-structured walls, the total friction includes the reaction forces of micro pillars due to flow shear and flow pressure at pillar surfaces and the reaction force of bottom plane due to flow shear on bottom surface. For larger Reynolds numbers, fluid inertia prevents the fluid from flowing along the curved surface of micro pillars and reduces the equivalent shear stress of the pillar reaction force due to flow shear. At the same time, the fluid inertia makes the overhead flow impact the windward side of micro pillars more strongly and therefore increases the equivalent shear stress of the pillar reaction force due to flow pressure.

physics.flu-dyn↗

arXiv2022

Enhancement of heat and mass transfer by herringbone microstructures in a simple shear flow

The heat and mass transfer characteristics in a simple shear flow over staggered herringbone structures are numerically investigated with the lattice Boltzmann method. Two flow motions are identified. The first is a spiral flow oscillation above the herringbone structures that advects heat and mass from the top plane to herringbone structures. The second is a flow recirculation in the grooves between herringbone ridges that advects heat and mass from the area around herringbone tips to the side walls of herringbone ridges and the bottom surfaces. These two basic flow motions couple together to form complex transport mechanisms. The results show that when advective heat and mass transfer takes effect at relatively larger Reynolds and Schmidt numbers, the dependence of the total transfer rate on the Schmidt number follows a power law, with the power being the same as that in the Dittus-Boelter equation for turbulent heat transfer. As Reynolds number increases, the dependence of the total transfer rate on Reynolds number also approaches a power law, and the power is close to that in the Dittus-Boelter equation.

physics.flu-dyn↗

arXiv2021

CondiDiag1.0: A flexible online diagnostic tool for conditional sampling and budget analysis in the E3SM atmosphere model (EAM)

Numerical models used in weather and climate prediction take into account a comprehensive set of atmospheric processes such as the resolved and unresolved fluid dynamics, radiative transfer, cloud and aerosol life cycles, and mass or energy exchanges with the Earth's surface. In order to identify model deficiencies and improve predictive skills, it is important to obtain process-level understanding of the interactions between different processes. Conditional sampling and budget analysis are powerful tools for process-oriented model evaluation, but they often require tedious ad hoc coding and large amounts of instantaneous model output, resulting in inefficient use of human and computing resources. This paper presents an online diagnostic tool that addresses this challenge by monitoring model variables in a generic manner as they evolve within the time integration cycle. The tool is convenient to use. It allows users to select sampling conditions and specify monitored variables at run time. Both the evolving values of the model variables and their increments caused by different atmospheric processes can be monitored and archived. Online calculation of vertical integrals is also supported. Multiple sampling conditions can be monitored in a single simulation in combination with unconditional sampling. The paper explains in detail the design and implementation of the tool in the Energy Exascale Earth System Model (E3SM) version 1. The usage is demonstrated through three examples: a global budget analysis of dust aerosol mass concentration, a composite analysis of sea salt emission and its dependency on surface wind speed, and a conditionally sampled relative humidity budget. The tool is expected to be easily portable to closely related atmospheric models that use the same or similar data structures and time integration methods.

physics.ao-ph↗

arXiv2021

MultiDoc2Dial: Modeling Dialogues Grounded in Multiple Documents

We propose MultiDoc2Dial, a new task and dataset on modeling goal-oriented dialogues grounded in multiple documents. Most previous works treat document-grounded dialogue modeling as a machine reading comprehension task based on a single given document or passage. In this work, we aim to address more realistic scenarios where a goal-oriented information-seeking conversation involves multiple topics, and hence is grounded on different documents. To facilitate such a task, we introduce a new dataset that contains dialogues grounded in multiple documents from four different domains. We also explore modeling the dialogue-based and document-based context in the dataset. We present strong baseline approaches and various experimental results, aiming to support further research efforts on such a task.

cs.CL↗

arXiv2021

Statistical computation methods for microbiome compositional data network inference

Microbes can affect processes from food production to human health. Such microbes are not isolated, but rather interact with each other and establish connections with their living environments. Understanding these interactions is essential to an understanding of the organization and complex interplay of microbial communities, as well as the structure and dynamics of various ecosystems. A common and essential approach toward this objective involves the inference of microbiome interaction networks. Although network inference methods in other fields have been studied before, applying these methods to estimate microbiome associations based on compositional data will not yield valid results. On the one hand, features of microbiome data such as compositionality, sparsity and high-dimensionality challenge the data normalization and the design of computational methods. On the other hand, several issues like microbial community heterogeneity, external environmental interference and biological concerns also make it more difficult to deal with the network inference. In this paper, we provide a comprehensive review of emerging microbiome interaction network inference methods. According to various assumptions and research targets, estimated networks are divided into four main categories: correlation networks, conditional correlation networks, mixture networks and differential networks. Their scope of applications, advantages and limitations are presented in this review. Since real microbial interactions can be complex and dynamic, no unifying method has captured all the aspects of interest to date. In addition, we discuss the challenges now confronting current microbial associations study and future prospects. Finally, we highlight that the research in microbial network inference requires the joint promotion of statistical computation methods and experimental techniques.

stat.AP↗

arXiv2021

Explaining Neural Network Predictions on Sentence Pairs via Learning Word-Group Masks

Explaining neural network models is important for increasing their trustworthiness in real-world applications. Most existing methods generate post-hoc explanations for neural network models by identifying individual feature attributions or detecting interactions between adjacent features. However, for models with text pairs as inputs (e.g., paraphrase identification), existing methods are not sufficient to capture feature interactions between two texts and their simple extension of computing all word-pair interactions between two texts is computationally inefficient. In this work, we propose the Group Mask (GMASK) method to implicitly detect word correlations by grouping correlated words from the input text pair together and measure their contribution to the corresponding NLP tasks as a whole. The proposed method is evaluated with two different model architectures (decomposable attention model and BERT) across four datasets, including natural language inference and paraphrase identification tasks. Experiments show the effectiveness of GMASK in providing faithful explanations to these models.

cs.CL↗

arXiv2021

Quantifying and attributing time step sensitivities in present-day climate simulations conducted with EAMv1

This study assesses the relative importance of time integration error in present-day climate simulations conducted with the atmosphere component of the Energy Exascale Earth System Model version 1 (EAMv1) at 1-degree horizontal resolution. We show that a factor-of-6 reduction of time step size in all major parts of the model leads to significant changes in the long-term mean climate. These changes imply that the reduction of temporal truncation errors leads to a notable although unsurprising degradation of agreement between the simulated and observed present-day climate; the model would require retuning to regain optimal climate fidelity in the absence of those truncation errors. A coarse-grained attribution of the time step sensitivities is carried out by separately shortening time steps used in various components of EAM or by revising the numerical coupling between some processes. The results provide useful clues to help better understand the root causes of time step sensitivities in EAM. The experimentation strategy used here can also provide a pathway for other models to identify and reduce time integration errors.

physics.ao-ph↗

arXiv2020

doc2dial: A Goal-Oriented Document-Grounded Dialogue Dataset

We introduce doc2dial, a new dataset of goal-oriented dialogues that are grounded in the associated documents. Inspired by how the authors compose documents for guiding end users, we first construct dialogue flows based on the content elements that corresponds to higher-level relations across text sections as well as lower-level relations between discourse units within a section. Then we present these dialogue flows to crowd contributors to create conversational utterances. The dataset includes about 4800 annotated conversations with an average of 14 turns that are grounded in over 480 documents from four domains. Compared to the prior document-grounded dialogue datasets, this dataset covers a variety of dialogue scenes in information-seeking conversations. For evaluating the versatility of the dataset, we introduce multiple dialogue modeling tasks and present baseline approaches.

cs.CL↗

arXiv2020

Multi-task Learning with Multi-head Attention for Multi-choice Reading Comprehension

Multiple-choice Machine Reading Comprehension (MRC) is an important and challenging Natural Language Understanding (NLU) task, in which a machine must choose the answer to a question from a set of choices, with the question placed in context of text passages or dialog. In the last a couple of years the NLU field has been revolutionized with the advent of models based on the Transformer architecture, which are pretrained on massive amounts of unsupervised data and then fine-tuned for various supervised learning NLU tasks. Transformer models have come to dominate a wide variety of leader-boards in the NLU field; in the area of MRC, the current state-of-the-art model on the DREAM dataset (see[Sunet al., 2019]) fine tunes Albert, a large pretrained Transformer-based model, and addition-ally combines it with an extra layer of multi-head attention between context and question-answer[Zhuet al., 2020].The purpose of this note is to document a new state-of-the-art result in the DREAM task, which is accomplished by, additionally, performing multi-task learning on two MRC multi-choice reading comprehension tasks (RACE and DREAM).

cs.CL↗

arXiv2019

Theory-Software Translation: Research Challenges and Future Directions

The Theory-Software Translation Workshop, held in New Orleans in February 2019, explored in depth the process of both instantiating theory in software - for example, implementing a mathematical model in code as part of a simulation - and using the outputs of software - such as the behavior of a simulation - to advance knowledge. As computation within research is now ubiquitous, the workshop provided a timely opportunity to reflect on the particular challenges of research software engineering - the process of developing and maintaining software for scientific discovery. In addition to the general challenges common to all software development projects, research software additionally must represent, manipulate, and provide data for complex theoretical constructs. Ensuring this process is robust is essential to maintaining the integrity of the science resulting from it, and the workshop highlighted a number of areas where the current approach to research software engineering would benefit from an evidence base that could be used to inform best practice. The workshop brought together expert research software engineers and academics to discuss the challenges of Theory-Software Translation over a two-day period. This report provides an overview of the workshop activities, and a synthesises of the discussion that was recorded. The body of the report presents a thematic analysis of the challenges of Theory-Software Translation as identified by workshop participants, summarises these into a set of research areas, and provides recommendations for the future direction of this work.

cs.SE↗

arXiv2019

Improving solution accuracy and convergence for stochastic physics parameterizations with colored noise

Stochastic parameterizations are used in numerical weather prediction and climate modeling to help capture the uncertainty in the simulations and improve their statistical properties. Convergence issues can arise when time integration methods originally developed for deterministic differential equations are applied naively to stochastic problems. (Hodyss et al 2013, 2014) demonstrated that a correction term to various deterministic numerical schemes, known in stochastic analysis as the Itô correction, can help improve solution accuracy and ensure convergence to the physically relevant solution without substantial computational overhead. The usual formulation of the Itô correction is valid only when the stochasticity is represented by {\it white} noise. In this study, a generalized formulation of the Itô correction is derived for noises of any color. The formulation is applied to a test problem described by an advection-diffusion equation forced with a spectrum of fast processes. We present numerical results for cases with both constant and spatially varying advection velocities to show that, for the same time step sizes, the introduction of the generalized Itô correction helps to substantially reduce time integration error and significantly improve the convergence rate of the numerical solutions when the forcing term in the governing equation is rough (fast varying); alternatively, for the same target accuracy, the generalized Itô correction allows for the use of significantly longer time steps and hence helps to reduce the computational cost of the numerical simulation.

math.NA↗

arXiv2019

Rewarding Smatch: Transition-Based AMR Parsing with Reinforcement Learning

Our work involves enriching the Stack-LSTM transition-based AMR parser (Ballesteros and Al-Onaizan, 2017) by augmenting training with Policy Learning and rewarding the Smatch score of sampled graphs. In addition, we also combined several AMR-to-text alignments with an attention mechanism and we supplemented the parser with pre-processed concept identification, named entities and contextualized embeddings. We achieve a highly competitive performance that is comparable to the best published results. We show an in-depth study ablating each of the new components of the parser

cs.CL↗

arXiv2017

Endo-reversible heat engines coupled to finite thermal reservoirs: A rigorous treatment

We consider two specific thermodynamic cycles of engine operating in a finite time coupled to two thermal reservoirs with a finite heat capacity: The Carnot-type cycle and the Lorenz-type cycle. By means of the endo-reversible thermodynamics, we then discuss the power output of engine and its optimization. In doing so, we treat the temporal duration of a single cycle rigorously, i.e., without neglecting the duration of its adiabatic parts. Then we find that the maximally obtainable power output P_m and the engine efficiency η_m at the point of P_m explicitly depend on the heat conductance and the compression ratio. From this, it is immediate to observe that the well-known results available in many references, in particular the (compression-ratio-independent) Curzon-Ahlborn-Novikov expressions such as η_m --> η_{CAN} = 1 - (T_L/T_H)^(1/2) with the temperatures (T_H, T_L) of hot and cold reservoirs only, can be recovered, but, significantly enough, in the limit of a vanishingly small heat conductance and an infinitely large compression ratio only. Our result implies that the endo-reversible model of a thermal machine operating in a finite time and so producing a finite power output with the Curzon-Ahlborn-Novikov results should be limited in its own validity regime.

physics.gen-ph↗

arXiv2017

Recent progress and review of issues related to Physics Dynamics Coupling in geophysical models

Geophysical models of the atmosphere and ocean invariably involve parameterizations. These represent two distinct areas: Subgrid processes that the model cannot resolve, and diabatic sources in the equations, due to radiation for example. Hence, coupling between these physics parameterizations and the resolved fluid dynamics and also between the dynamics of the air and water, is necessary. In this paper weather and climate models are used to illustrate the problems. Nevertheless the same applies to other geophysical models. This coupling is an important aspect of geophysical models. However, often model development is strictly segregated into either physics or dynamics. As a consequence, this area has many unanswered questions. Recent developments in the design of dynamical cores, extended process physics and predicted future changes of the computational infrastructure are increasing complexity. This paper reviews the state-of-the-art of the physics-dynamics coupling in geophysical models, surveys the analysis techniques, and illustrates open questions in this field. This paper focuses on two objectives: To illustrate the phenomenology of the coupling problem with references to examples in the literature and to show how the problem can be analysed. Proposals are made on how to advance the understanding and upcoming challenges with emerging modeling strategies. This paper is of interest to model developers who aim to improve the models and have to make choices on and test new implementations, to users who have to understand choices presented to them and finally users of outputs, who have to distinguish physical features from numerical problems in the model data.

physics.ao-ph↗

arXiv2016

A Proximity Measure using Blink Model

This paper proposes a new graph proximity measure. This measure is a derivative of network reliability. By analyzing its properties and comparing it against other proximity measures through graph examples, we demonstrate that it is more consistent with human intuition than competitors. A new deterministic algorithm is developed to approximate this measure with practical complexity. Empirical evaluation by two link prediction benchmarks, one in coauthorship networks and one in Wikipedia, shows promising results. For example, a single parameterization of this measure achieves accuracies that are 14-35% above the best accuracy for each graph of all predictors reported in the 2007 Liben-Nowell and Kleinberg survey.

cs.SI↗

arXiv2011

Cicada: a Heavy but Agile Flyer

"Cicada: a Heavy but Agile Flyer" is a fluid dynamic video submitted to Gallery of Fluid Motion in APS-DFD 2011. Comparing to other insects, cicadas can generate much higher lift to overcome their large body weight. The hidden mechanism may help in designing a Micro Air Vehicle (MAV) to carry large payloads. However, it is lack of literatures in discussing how cicadas use their wings to accomplish various flights. In this work, a high-speed photogrammetry system and 3D surface reconstruction technology are used to reveal cicada wing kinematics and deformation during a freely forward flight. The aerodynamic performance is studied using in-house immerse boundary method based Computational Fluid Dynamics(CFD) solver.

physics.flu-dyn↗