arXiv subjects

Hai Wang

Publications and source records attributed to Hai Wang.

At least 91 records · Page 5Linked to original sources

arXiv2020

Semi-Analytical Model for Design and Analysis of On-Orbit Servicing Architecture

Robotic on-orbit servicing (OOS) is expected to be a key technology and concept for future sustainable space exploration. This paper develops a semi-analytical model for OOS systems analysis, responding to the growing needs and ongoing trend of robotic OOS. An OOS infrastructure system is considered whose goal is to provide responsive services to the random failures of a set of customer modular satellites distributed in space (e.g., at the geosynchronous equatorial orbit). The considered OOS architecture is comprised of a servicer that travels and provides module-replacement services to the customer satellites, an on-orbit depot to store the spares, and a series of launch vehicles to replenish the depot. The OOS system performance is analyzed by evaluating the mean waiting time before service completion for a given failure and its relationship with the depot capacity. Leveraging the queueing theory and inventory management methods, the developed semi-analytical model is capable of analyzing the OOS system performance without relying on computationally costly simulations. The effectiveness of the proposed model is demonstrated using a case study compared with simulation results. This paper is expected to provide a critical step to push the research frontier of analytical/semi-analytical models development for complex space systems design.

cs.PF↗

arXiv2020

Magnon-magnon interaction and magnon relaxation time in ferromagnetic Cr2Ge2Te6 monolayer

Despite the intense amount of attention and huge potential of two-dimensional (2D) magnets for applications in novel magnetic, magneto-optical, magneto-thermal and magneto-electronic devices, there has yet to be a robust strategy developed to systematically understand magnon-magnon (MMI) interactions at finite temperature. In this paper, we present a first-principles theoretical method to introduce the finite temperature magnon-magnon interaction into Heisenberg Hamiltonian through a nonlinear correction energy. The Wick theorem is used to decouple the four-magnon operators to two-magnon order. We demonstrate the capabilities of this method by studying the strength of MMI in Cr2Ge2Te6 (CGT) monolayer. The spin wave spectrum at finite temperature and the time-dependent spin autocorrelation function are explored. It is found that the magnon relaxation time due to magnon-magnon scattering increases with temperature because of the reduction in magnon energy, while decreases with wavevector and external magnetic field. Our results provide a new insight to understand the magnon damping and energy dissipation in two-dimensional ferromagnetic materials.

cond-mat.mes-hall↗

arXiv2020

Knowledge Efficient Deep Learning for Natural Language Processing

Deep learning has become the workhorse for a wide range of natural language processing applications. But much of the success of deep learning relies on annotated examples. Annotation is time-consuming and expensive to produce at scale. Here we are interested in methods for reducing the required quantity of annotated data -- by making the learning methods more knowledge efficient so as to make them more applicable in low annotation (low resource) settings. There are various classical approaches to making the models more knowledge efficient such as multi-task learning, transfer learning, weakly supervised and unsupervised learning etc. This thesis focuses on adapting such classical methods to modern deep learning models and algorithms. This thesis describes four works aimed at making machine learning models more knowledge efficient. First, we propose a knowledge rich deep learning model (KRDL) as a unifying learning framework for incorporating prior knowledge into deep models. In particular, we apply KRDL built on Markov logic networks to denoise weak supervision. Second, we apply a KRDL model to assist the machine reading models to find the correct evidence sentences that can support their decision. Third, we investigate the knowledge transfer techniques in multilingual setting, where we proposed a method that can improve pre-trained multilingual BERT based on the bilingual dictionary. Fourth, we present an episodic memory network for language modelling, in which we encode the large external knowledge for the pre-trained GPT.

cs.CL↗

arXiv2020

Competitive ride-sourcing market with a third-party integrator

Recently, some transportation service providers attempt to integrate the ride services offered by multiple independent ride-sourcing platforms, and passengers are able to request ride through such third-party integrators or connectors and receive service from any one of the platforms. This novel business model, termed as third-party platform-integration in this paper, has potentials to alleviate the cost of market fragmentation due to the demand splitting among multiple platforms. While most existing studies focus on the operation strategies for one single monopolist platform, much less is known about the competition and platform-integration as well as the implications on operation strategy and system efficiency. In this paper, we propose mathematical models to describe the ride-sourcing market with multiple competing platforms and compare system performance metrics between two market scenarios, i.e., with and without platform-integration, at Nash equilibrium as well as social optimum. We find that platform-integration can increase total realized demand and social welfare at both Nash equilibrium and social optimum, but may not necessarily generate a greater profit when vehicle supply is sufficiently large or/and market is too fragmented. We show that the market with platform-integration generally achieves greater social welfare. On one hand, the integrator in platform-integration is able to generate a thicker market and reduce matching frictions; on the other hand, multiple platforms are still competing by independently setting their prices, which help to mitigate monopoly mark-up as in the monopoly market.

econ.GN↗

arXiv2020

On-The-Fly Information Retrieval Augmentation for Language Models

Here we experiment with the use of information retrieval as an augmentation for pre-trained language models. The text corpus used in information retrieval can be viewed as form of episodic memory which grows over time. By augmenting GPT 2.0 with information retrieval we achieve a zero shot 15% relative reduction in perplexity on Gigaword corpus without any re-training. We also validate our IR augmentation on an event co-reference task.

cs.CL↗

arXiv2020

Long-lived and multiplexed atom-photon entanglement interface with feed-forward-controlled readouts

The quantum interface (QI) that generates entanglement between photonic and spin-wave (atomic memory) qubits is a basic building block for quantum repeaters. Realizing ensemble-based repeaters in practice requires quantum memory providing long lifetime and multimode capacity. Significant progresses have been achieved on these separate goals. The remaining challenge is to combine long-lived and multimode memories into a single QI. Here, by establishing multimode, magnetic-field-insensitive and long-wavelength spin-wave storage in laser-cooled atoms that are placed inside a phase-passively-stabilized polarization interferometer, we constructed a multiplexed QI that stores up to three long-lived spin-wave qubits. Using a feed-forward-controlled system, we demonstrated that the multiplexed QI gives rise to a 3-fold increase in the atom-photon (photon-photon) entanglement-generation probability compared to single-mode QIs. The measured Bell parameter is 2.5+/-0.1 combined with a memory lifetime up to 1ms. The presented work represents a key step forward in realizing fiber-based long-distance quantum communications.

quant-ph↗

arXiv2020

MixPUL: Consistency-based Augmentation for Positive and Unlabeled Learning

Learning from positive and unlabeled data (PU learning) is prevalent in practical applications where only a couple of examples are positively labeled. Previous PU learning studies typically rely on existing samples such that the data distribution is not extensively explored. In this work, we propose a simple yet effective data augmentation method, coined~\algo, based on \emph{consistency regularization} which provides a new perspective of using PU data. In particular, the proposed~\algo~incorporates supervised and unsupervised consistency training to generate augmented data. To facilitate supervised consistency, reliable negative examples are mined from unlabeled data due to the absence of negative samples. Unsupervised consistency is further encouraged between unlabeled datapoints. In addition,~\algo~reduces margin loss between positive and unlabeled pairs, which explicitly optimizes AUC and yields faster convergence. Finally, we conduct a series of studies to demonstrate the effectiveness of consistency regularization. We examined three kinds of reliable negative mining methods. We show that~\algo~achieves an averaged improvement of classification error from 16.49 to 13.09 on the CIFAR-10 dataset across different positive data amount.

cs.LG↗

arXiv2019

Improving Pre-Trained Multilingual Models with Vocabulary Expansion

Recently, pre-trained language models have achieved remarkable success in a broad range of natural language processing tasks. However, in multilingual setting, it is extremely resource-consuming to pre-train a deep language model over large-scale corpora for each language. Instead of exhaustively pre-training monolingual language models independently, an alternative solution is to pre-train a powerful multilingual deep language model over large-scale corpora in hundreds of languages. However, the vocabulary size for each language in such a model is relatively small, especially for low-resource languages. This limitation inevitably hinders the performance of these multilingual models on tasks such as sequence labeling, wherein in-depth token-level or sentence-level understanding is essential. In this paper, inspired by previous methods designed for monolingual settings, we investigate two approaches (i.e., joint mapping and mixture mapping) based on a pre-trained multilingual model BERT for addressing the out-of-vocabulary (OOV) problem on a variety of tasks, including part-of-speech tagging, named entity recognition, machine translation quality estimation, and machine reading comprehension. Experimental results show that using mixture mapping is more promising. To the best of our knowledge, this is the first work that attempts to address and discuss the OOV issue in multilingual settings.

cs.CL↗

arXiv2019

Evidence Sentence Extraction for Machine Reading Comprehension

Remarkable success has been achieved in the last few years on some limited machine reading comprehension (MRC) tasks. However, it is still difficult to interpret the predictions of existing MRC models. In this paper, we focus on extracting evidence sentences that can explain or support the answers of multiple-choice MRC tasks, where the majority of answer options cannot be directly extracted from reference documents. Due to the lack of ground truth evidence sentence labels in most cases, we apply distant supervision to generate imperfect labels and then use them to train an evidence sentence extractor. To denoise the noisy labels, we apply a recently proposed deep probabilistic logic learning framework to incorporate both sentence-level and cross-sentence linguistic indicators for indirect supervision. We feed the extracted evidence sentences into existing MRC models and evaluate the end-to-end performance on three challenging multiple-choice MRC datasets: MultiRC, RACE, and DREAM, achieving comparable or better performance than the same models that take as input the full reference document. To the best of our knowledge, this is the first work extracting evidence sentences for multiple-choice MRC.

cs.CL↗

arXiv2019

Deterministic generation and partial retrieval of a spin-wave excitation in an atomic ensemble

In this paper, we report a generation of a spin-wave excitation (SWE) with a near-unity () probability in a given time (~730). Such deterministic generation relies on a feedback scheme with a millisecond quantum memory. The millisecond memory is achieved by maximizing the wavelength of the spin wave and storing the SWE as the magnetic-field-insensitive transition. We then demonstrate partial retrievals of the spin wave by applying a first read pulse whose area is smaller than the value of . The remained SWE is fully retrieved by a second pulse. Anti-correlation function between the detections in the first and second readouts has been measured, which shows that the partial-retrieval operation on the SWE is in the quantum regime. The presented experiment represents an important step towards the realization of the improved DLCZ quantum repeater protocol proposed in Phys. Rev. A 77, 062301 (2008).

quant-ph↗

arXiv2019

Discrete-Time Modelling of Quantum Evolutions, the Energy-Time Uncertainty Relation and General Extensions in the Entangled History Formalism

Time evolution is an indivisible part in any physics theory. Usually, people are accustomed to think that the universe is a fixed background and the system itself evolves step by step in time. However, Yakir Aharonov challenges this view using his two-vector formalism. In this paper, using the entangled history formalism, we attain three achievements.

quant-ph↗

arXiv2019

Multiplexed spin-wave-photon entanglement source using temporal-multimode memories and feedforward-controlled readout

The sources, which generate atom-photon quantum correlations or entanglement based on quantum memory, are basic blocks for building quantum repeaters (QRs). For achieving highly entanglement-generation rates in ensemble-based QRs, spatial-, temporal- and spectral-multimode memories are needed. The previous temporal-multimode memories are based on rephrasing mechanisms in inhomogeneous-broadened media. Here, by applying a train of multi-direction write pulses into a homogeneous-broadened atomic ensemble to induce Duan-Lukin-Cirac-Zoller-like Raman processes, we prepare up to 19 pairs of modes, namely, a spin-wave mode and a photonic time bin. Spin-wave-photon (i.e., atom-photon) entanglement is probabilistically produced in the mode pairs. As a proof-in-principle demonstration, we show that the multiplexed source using all the spin-wave modes and feed-forward-control readout increases spin-wave-photon entanglement generation rate by a factor of $\sim$18.3, compared to non-multiplexed source using the individual modes. The measured Bell parameter for the multiplexed source is $2.30\pm 0.02$ combined with a memory lifetime of $30μs$.

quant-ph↗

arXiv2018

To Compress, or Not to Compress: Characterizing Deep Learning Model Compression for Embedded Inference

The recent advances in deep neural networks (DNNs) make them attractive for embedded systems. However, it can take a long time for DNNs to make an inference on resource-constrained computing devices. Model compression techniques can address the computation issue of deep inference on embedded devices. This technique is highly attractive, as it does not rely on specialized hardware, or computation-offloading that is often infeasible due to privacy concerns or high latency. However, it remains unclear how model compression techniques perform across a wide range of DNNs. To design efficient embedded deep learning solutions, we need to understand their behaviors. This work develops a quantitative approach to characterize model compression techniques on a representative embedded deep learning architecture, the NVIDIA Jetson Tx2. We perform extensive experiments by considering 11 influential neural network architectures from the image classification and the natural language processing domains. We experimentally show that how two mainstream compression techniques, data quantization and pruning, perform on these network architectures and the implications of compression techniques to the model storage size, inference time, energy consumption and performance metrics. We demonstrate that there are opportunities to achieve fast deep inference on embedded systems, but one must carefully choose the compression settings. Our results provide insights on when and how to apply model compression techniques and guidelines for designing efficient embedded deep learning systems.

cs.LG↗

arXiv2018

Learning to Globally Edit Images with Textual Description

We show how we can globally edit images using textual instructions: given a source image and a textual instruction for the edit, generate a new image transformed under this instruction. To tackle this novel problem, we develop three different trainable models based on RNN and Generative Adversarial Network (GAN). The models (bucket, filter bank, and end-to-end) differ in how much expert knowledge is encoded, with the most general version being purely end-to-end. To train these systems, we use Amazon Mechanical Turk to collect textual descriptions for around 2000 image pairs sampled from several datasets. Experimental results evaluated on our dataset validate our approaches. In addition, given that the filter bank model is a good compromise between generality and performance, we investigate it further by replacing RNN with Graph RNN, and show that Graph RNN improves performance. To the best of our knowledge, this is the first computational photography work on global image editing that is purely based on free-form textual instructions.

cs.CV↗

arXiv2018

Deep Probabilistic Logic: A Unifying Framework for Indirect Supervision

Deep learning has emerged as a versatile tool for a wide range of NLP tasks, due to its superior capacity in representation learning. But its applicability is limited by the reliance on annotated examples, which are difficult to produce at scale. Indirect supervision has emerged as a promising direction to address this bottleneck, either by introducing labeling functions to automatically generate noisy examples from unlabeled text, or by imposing constraints over interdependent label decisions. A plethora of methods have been proposed, each with respective strengths and limitations. Probabilistic logic offers a unifying language to represent indirect supervision, but end-to-end modeling with probabilistic logic is often infeasible due to intractable inference and learning. In this paper, we propose deep probabilistic logic (DPL) as a general framework for indirect supervision, by composing probabilistic logic with deep learning. DPL models label decisions as latent variables, represents prior knowledge on their relations using weighted first-order logical formulas, and alternates between learning a deep neural network for the end task and refining uncertain formula weights for indirect supervision, using variational EM. This framework subsumes prior indirect supervision methods as special cases, and enables novel combination via infusion of rich domain and linguistic knowledge. Experiments on biomedical machine reading demonstrate the promise of this approach.

cs.CL↗

arXiv2018

Self-Organizing Relay Selection in UAV Communication Networks: A Matching Game Perspective

For large unmanned aerial vehicle (UAV) networks, the timely communication is needed to accomplish a series of missions accurately and effectively. The relay technology will play an important role in UAV networks by helping drones communicating with long-distance drones, which solves the problem of the limited transmission power of drones. In this paper, the relay selection is seen as the entry point to improve the performance of self-organizing network with multiple optimizing factors. Different from the ground relay models, the relay selection in UAV communication networks presents new challenges, including heterogeneous, dynamic, dense and limited information characteristics. More effective schemes with distributed, fast, robust and scalable features are required to solve the optimizing problem. After discussing the challenges and requirements, we find that the matching game is suitable to model the complex relay model. The advantages of the matching game in self-organizing UAV communications are discussed. Moreover, we provide extensive applications of matching markets, and then propose a novel classification of matching game which focuses on the competitive relationship between players. Specifically, basic preliminary models are presented and some future research directions of matching game in UAV relay models are discussed.

cs.NI↗

arXiv2018

Field extension of real values of physical observables in classical theory can help attain quantum results

Physical quantities are assumed to take real values, which stems from the fact that an usual measuring instrument that measures a physical observable always yields a real number. Here we consider the question of what will happen if physical observables are allowed to take complex values. In this paper, we show that by allowing observables in the Bell inequality to take complex values, a classical physical theory can actually get the same upper bound of the Bell expression as quantum theory. Also, by extending the real field to the quaternionic field, we can puzzle out the GHZ problem using local hidden variable model. Furthermore, we try to build a new type of hidden-variable theory of a single qubit based on the result.

quant-ph↗

arXiv2017

High-quality trapped modes in all-dielectric metamaterials

A planar all-dielectric metamaterial made of a double-periodic lattice whose unit cell consists of a single subwavelength dielectric particle having the form of a disk possessing a penetrating hole is considered. The resonant states in the transmitted spectra of the metamaterial are identified considering modes inherent to the individual cylindrical dielectric resonator. A correlation between the asymmetry in particle's geometry, which arises from the off-centered displacement of the hole, and formation of the Mie-type and trapped modes is established. The advantages of using a coaxial-sector notch instead of a round hole for the trapped mode excitation are explained.

physics.class-ph↗