Searcharxiv⌕ Search

arXiv subjects

Di Wu

Publications and source records attributed to Di Wu.

At least 577 records · Page 32Linked to original sources
arXiv2022

Analytical Shaping Method for Low-Thrust Rendezvous Trajectory Using Cubic Spline Functions

Preliminary mission design requires an efficient and accurate approximation to the low-thrust rendezvous trajectories, which might be generally three-dimensional and involve multiple revolutions. In this paper, a new shaping method using cubic spline functions is developed for the analytical approximation, which shows advantages in the optimality and computational efficiency. The rendezvous constraints on the boundary states and transfer time are all satisfied analytically, under the assumption that the boundary conditions and segment numbers of cubic spline functions are designated in advance. Two specific shapes are then formulated according to whether they have free optimization parameters. The shape without free parameters provides an efficient and robust estimation, while the other one allows a subsequent optimization for the satisfaction of additional constraints such as the constraint on the thrust magnitude. Applications of the proposed method in combination with the particle swarm optimization algorithm are discussed through two typical interplanetary rendezvous missions, that is, an inclined multi-revolution trajectory from the Earth to asteroid Dionysus and a multi-rendezvous trajectory of sample return. Simulation examples show that the proposed method is superior to existing methods in terms of providing good estimation for the global search and generating suitable initial guess for the subsequent trajectory optimization.

cs.RO↗
arXiv2021

U2++: Unified Two-pass Bidirectional End-to-end Model for Speech Recognition

The unified streaming and non-streaming two-pass (U2) end-to-end model for speech recognition has shown great performance in terms of streaming capability, accuracy, real-time factor (RTF), and latency. In this paper, we present U2++, an enhanced version of U2 to further improve the accuracy. The core idea of U2++ is to use the forward and the backward information of the labeling sequences at the same time at training to learn richer information, and combine the forward and backward prediction at decoding to give more accurate recognition results. We also proposed a new data augmentation method called SpecSub to help the U2++ model to be more accurate and robust. Our experiments show that, compared with U2, U2++ shows faster convergence at training, better robustness to the decoding method, as well as consistent 5\% - 8\% word error rate reduction gain over U2. On the experiment of AISHELL-1, we achieve a 4.63\% character error rate (CER) with a non-streaming setup and 5.05\% with a streaming setup with 320ms latency by U2++. To the best of our knowledge, 5.05\% is the best-published streaming result on the AISHELL-1 test set.

cs.SD↗
arXiv2021

Unified Streaming and Non-streaming Two-pass End-to-end Model for Speech Recognition

In this paper, we present a novel two-pass approach to unify streaming and non-streaming end-to-end (E2E) speech recognition in a single model. Our model adopts the hybrid CTC/attention architecture, in which the conformer layers in the encoder are modified. We propose a dynamic chunk-based attention strategy to allow arbitrary right context length. At inference time, the CTC decoder generates n-best hypotheses in a streaming way. The inference latency could be easily controlled by only changing the chunk size. The CTC hypotheses are then rescored by the attention decoder to get the final result. This efficient rescoring process causes very little sentence-level latency. Our experiments on the open 170-hour AISHELL-1 dataset show that, the proposed method can unify the streaming and non-streaming model simply and efficiently. On the AISHELL-1 test set, our unified model achieves 5.60% relative character error rate (CER) reduction in non-streaming ASR compared to a standard non-streaming transformer. The same model achieves 5.42% CER with 640ms latency in a streaming ASR system.

cs.SD↗
arXiv2021

WeNet: Production oriented Streaming and Non-streaming End-to-End Speech Recognition Toolkit

In this paper, we propose an open source, production first, and production ready speech recognition toolkit called WeNet in which a new two-pass approach is implemented to unify streaming and non-streaming end-to-end (E2E) speech recognition in a single model. The main motivation of WeNet is to close the gap between the research and the production of E2E speechrecognition models. WeNet provides an efficient way to ship ASR applications in several real-world scenarios, which is the main difference and advantage to other open source E2E speech recognition toolkits. In our toolkit, a new two-pass method is implemented. Our method propose a dynamic chunk-based attention strategy of the the transformer layers to allow arbitrary right context length modifies in hybrid CTC/attention architecture. The inference latency could be easily controlled by only changing the chunk size. The CTC hypotheses are then rescored by the attention decoder to get the final result. Our experiments on the AISHELL-1 dataset using WeNet show that, our model achieves 5.03\% relative character error rate (CER) reduction in non-streaming ASR compared to a standard non-streaming transformer. After model quantification, our model perform reasonable RTF and latency.

cs.SD↗
arXiv2021

Shadowless rapidly rotating yet not ultraspinning Kerr-AdS$_4$ and Kerr-Newman-AdS$_4$ black holes

We find that the Kerr-(Newman)-AdS$_4$ black hole will be shadowless if its rotation parameter is larger than a critical value and the shadowless-ness may be related to the appearance of the null hypersurface caustics (NHC) both inside the Cauchy horizon and outside the event horizon for the black hole with the rotation parameter beyond the critical value. Our studies also further confirm that whether an ultraspinning black hole is super-entropic or not is unrelated to the existence of the NHC outside the event horizon.

gr-qc↗
arXiv2021

ModelLight: Model-Based Meta-Reinforcement Learning for Traffic Signal Control

Traffic signal control is of critical importance for the effective use of transportation infrastructures. The rapid increase of vehicle traffic and changes in traffic patterns make traffic signal control more and more challenging. Reinforcement Learning (RL)-based algorithms have demonstrated their potential in dealing with traffic signal control. However, most existing solutions require a large amount of training data, which is unacceptable for many real-world scenarios. This paper proposes a novel model-based meta-reinforcement learning framework (ModelLight) for traffic signal control. Within ModelLight, an ensemble of models for road intersections and the optimization-based meta-learning method are used to improve the data efficiency of an RL-based traffic light control method. Experiments on real-world datasets demonstrate that ModelLight can outperform state-of-the-art traffic light control algorithms while substantially reducing the number of required interactions with the real-world environment.

cs.LG↗
arXiv2021

Gravitational wave echoes search with combs

Gravitational wave echoes may provide a smoking gun signal for new physics in the immediate vicinity of black holes. As a quasiperiodic signal in time, echoes are characterized by the nearly constant time delay, and its precise measurement can help reveal a Planck-scale deviation outside of the would-be horizon. Different search methods have been developed for this quasiperiodic signal, while the searches suffer from large theoretical uncertainties of the echo waveform associated with the near-horizon physics. On the other hand, a coherent combination of a large number of pulses gives rise to a generic narrow resonance structure for the echo amplitude in frequency. The quasiperiodic resonance structure sets a complementary search target for echoes, and the time delay is inversely related to the average resonance spacing. A uniform comb has been proposed to look for the resonance structure in a rather model-independent way. In this paper, we develop a Bayesian algorithm to search for the resonance structure based on combs, where a phase-marginalized likelihood plays an essential role. The algorithm is validated with signal injections in detector noise from Advanced LIGO. With special treatments of the non-Gaussian artifacts, the noise outliers of the log Bayes factor distribution are properly removed. An echo signal not significantly below noise is detectable, and the time delay can be determined to very high precision. We perform the proposed search on real gravitational wave strain data of the first observing run of Advanced LIGO. We find no clear evidence of a comblike structure for GW150914 and GW151012.

gr-qc↗
arXiv2021

Environment-Aware Beam Selection for IRS-Aided Communication with Channel Knowledge Map

Intelligent reflecting surface (IRS)-aided communication is a promising technology for beyond 5G (B5G) systems, to reconfigure the radio environment proactively. However, IRS-aided communication in practice requires efficient channel estimation or passive beam training, whose overhead and complexity increase drastically with the number of reflecting elements/beam directions. To tackle this challenge, we propose in this paper a novel environment-aware joint active and passive beam selection scheme for IRS-aided wireless communication, based on the new concept of channel knowledge map (CKM). Specifically, by utilizing both the location information of the user equipment (UE), which is readily available in contemporary wireless systems with ever-increasing accuracy, and the environment information offered by CKM, the proposed scheme achieves efficient beam selection with either no real-time training required (training-free beam selection) or only moderate training overhead (light-training beam selection). Numerical results based on practical channels obtained using commercial ray tracing software are presented, which demonstrate the superior performance of the proposed scheme over various benchmark schemes.

cs.IT↗
arXiv2021

Transition of laser-induced terahertz spin currents from torque- to conduction-electron-mediated transport

Spin transport is crucial for future spintronic devices operating at bandwidths up to the terahertz (THz) range. In F|N thin-film stacks made of a ferro/ferrimagnetic layer F and a normal-metal layer N, spin transport is mediated by (1) spin-polarized conduction electrons and/or (2) torque between electron spins. To identify a cross-over from (1) to (2), we study laser-driven spin currents in F|Pt stacks where F consists of model materials with different degrees of electrical conductivity. For the magnetic insulators YIG, GIG and maghemite, identical dynamics is observed. It arises from the THz interfacial spin Seebeck effect (SSE), is fully determined by the relaxation of the electrons in the metal layer and provides an estimate of the spin-mixing conductance of the GIG/Pt interface. Remarkably, in the half-metallic ferrimagnet Fe3O4 (magnetite), our measurements reveal two spin-current components with opposite direction. The slower, positive component exhibits SSE dynamics and is assigned to torque-type magnon excitation of the A- and B-spin sublattices of Fe3O4. The faster, negative component arises from the pyro-spintronic effect and can consistently be assigned to ultrafast demagnetization of e-sublattice minority-spin hopping electrons. This observation supports the magneto-electronic model of Fe3O4. In general, our results provide a new route to the contact-free separation of torque- and conduction-electron-mediated spin currents.

cond-mat.mes-hall↗
arXiv2021

Ultra-spinning Chow's black holes in six-dimensional gauged supergravity and their properties

By taking the ultra-spinning limit as a simple solution-generating trick, a novel class of ultra-spinning charged black hole solutions has been constructed from Chow's rotating charged black hole with two equal-charge parameters in six-dimensional $\mathcal{N} = 4$ gauged supergravity theory. We investigate their thermodynamical properties and then demonstrate that all thermodynamical quantities completely obey both the differential first law and the Bekenstein-Smarr mass formula. For the six-dimensional ultra-spinning Chow's black hole with only one rotation parameter, we show that it does not always obey the reverse isoperimetric inequality, thus it can be either sub-entropic or super-entropic, depending upon the ranges of the mass parameter and especially the charge parameter. This property is obviously different from that of the six-dimensional singly-rotating Kerr-AdS super-entropic black hole, which always strictly violates the RII. For the six-dimensional doubly-rotating Chow's black hole but ultra-spinning only along one spatial axis, we point out that it may also obey or violate the RII, and can be either super-entropic or sub-entropic in general.

hep-th↗
arXiv2021

Improving Neural Machine Translation by Bidirectional Training

We present a simple and effective pretraining strategy -- bidirectional training (BiT) for neural machine translation. Specifically, we bidirectionally update the model parameters at the early stage and then tune the model normally. To achieve bidirectional updating, we simply reconstruct the training samples from "src$\rightarrow$tgt" to "src+tgt$\rightarrow$tgt+src" without any complicated model modifications. Notably, our approach does not increase any parameters or training steps, requiring the parallel data merely. Experimental results show that BiT pushes the SOTA neural machine translation performance across 15 translation tasks on 8 language pairs (data sizes range from 160K to 38M) significantly higher. Encouragingly, our proposed model can complement existing data manipulation strategies, i.e. back translation, data distillation, and data diversification. Extensive analyses show that our approach functions as a novel bilingual code-switcher, obtaining better bilingual alignment.

cs.CL↗
arXiv2021

Poisson metrics and Higgs bundles over noncompact Kähler manifolds

In this paper, we study the existence of Poisson metrics on flat vector bundles over noncompact Riemannian manifolds and discuss related consequence, specially on the applications in Higgs bundles, towards generalizing Corlette-Donaldson-Hitchin-Simpson's nonabelian Hodge correspondence to noncompact Kähler manifolds setting.

math.DG↗
arXiv2021

The USYD-JD Speech Translation System for IWSLT 2021

This paper describes the University of Sydney& JD's joint submission of the IWSLT 2021 low resource speech translation task. We participated in the Swahili-English direction and got the best scareBLEU (25.3) score among all the participants. Our constrained system is based on a pipeline framework, i.e. ASR and NMT. We trained our models with the officially provided ASR and MT datasets. The ASR system is based on the open-sourced tool Kaldi and this work mainly explores how to make the most of the NMT models. To reduce the punctuation errors generated by the ASR model, we employ our previous work SlotRefine to train a punctuation correction model. To achieve better translation performance, we explored the most recent effective strategies, including back translation, knowledge distillation, multi-feature reranking and transductive finetuning. For model structure, we tried auto-regressive and non-autoregressive models, respectively. In addition, we proposed two novel pre-train approaches, i.e. \textit{de-noising training} and \textit{bidirectional training} to fully exploit the data. Extensive experiments show that adding the above techniques consistently improves the BLEU scores, and the final submission system outperforms the baseline (Transformer ensemble model trained with the original parallel data) by approximately 10.8 BLEU score, achieving the SOTA performance.

cs.CL↗
arXiv2021

Time Series Anomaly Detection for Smart Grids: A Survey

With the rapid increase in the integration of renewable energy generation and the wide adoption of various electric appliances, power grids are now faced with more and more challenges. One prominent challenge is to implement efficient anomaly detection for different types of anomalous behaviors within power grids. These anomalous behaviors might be induced by unusual consumption patterns of the users, faulty grid infrastructures, outages, external cyberattacks, or energy fraud. Identifying such anomalies is of critical importance for the reliable and efficient operation of modern power grids. Various methods have been proposed for anomaly detection on power grid time-series data. This paper presents a short survey of the recent advances in anomaly detection for power grid time-series data. Specifically, we first outline current research challenges in the power grid anomaly detection domain and further review the major anomaly detection approaches. Finally, we conclude the survey by identifying the potential directions for future research.

cs.LG↗
arXiv2021

Optimizing the Numbers of Queries and Replies in Federated Learning with Differential Privacy

Federated learning (FL) empowers distributed clients to collaboratively train a shared machine learning model through exchanging parameter information. Despite the fact that FL can protect clients' raw data, malicious users can still crack original data with disclosed parameters. To amend this flaw, differential privacy (DP) is incorporated into FL clients to disturb original parameters, which however can significantly impair the accuracy of the trained model. In this work, we study a crucial question which has been vastly overlooked by existing works: what are the optimal numbers of queries and replies in FL with DP so that the final model accuracy is maximized. In FL, the parameter server (PS) needs to query participating clients for multiple global iterations to complete training. Each client responds a query from the PS by conducting a local iteration. Our work investigates how many times the PS should query clients and how many times each client should reply the PS. We investigate two most extensively used DP mechanisms (i.e., the Laplace mechanism and Gaussian mechanisms). Through conducting convergence rate analysis, we can determine the optimal numbers of queries and replies in FL with DP so that the final model accuracy can be maximized. Finally, extensive experiments are conducted with publicly available datasets: MNIST and FEMNIST, to verify our analysis and the results demonstrate that properly setting the numbers of queries and replies can significantly improve the final model accuracy in FL with DP.

cs.LG↗
arXiv2021

Data-driven Model Predictive and Reinforcement Learning Based Control for Building Energy Management: a Survey

Building energy management is one of the core problems in modern power grids to reduce energy consumption while ensuring occupants' comfort. However, the building energy management system (BEMS) is now facing more challenges and uncertainties with the increasing penetration of renewables and complicated interactions between humans and buildings. Classical model predictive control (MPC) has shown its capacity to reduce building energy consumption, but it suffers from labor-intensive modelling and complex on-line control optimization. Recently, with the growing accessibility to the building control and automation data, data-driven solutions have attracted more research interest. This paper presents a compact review of the recent advances in data-driven MPC and reinforcement learning based control methods for BEMS. The main challenges in these approaches and insights on the selection of a control method are discussed.

eess.SY↗
arXiv2021

Design of Small Multi-band Full-screen Smartwatch Antenna for IoT applications

Smartwatch is a potential candidate for the Internet of Things (IoT) hub. However, the performance of smartwatch antennas is severely restricted by the smartwatch structure, especially when the antennas are designed by traditional methods. For adapting smartwatches to the role of IoT hub, a novel method of designing multi-band smartwatch antenna is presented in this paper, aiming at increasing the number of frequency bands, omni-directivity, and structural suitability. Firstly, the fundamental structure (including the full screen and the system PCB) of the smartwatch is analyzed as a whole by characteristic mode analysis (CMA). Thus, abundant resources of characteristic modes are introduced. The fundamental structure is then modified as the radiator of a multi-band antenna. Then, a non-radiating capacitive coupling element (CCE) excites the desired four 0.5-wavelength modes from this structure. This method could fully utilize the intrinsic modes of the smartwatch structure itself, thus exhibits multiple advantages: significantly small size, smaller ground, omni-directional radiation, and fitting to the full-screen smartwatch structure.

physics.app-ph↗