arXiv subjects

Hongyu Li

Publications and source records attributed to Hongyu Li.

At least 145 records · Page 8Linked to original sources

arXiv2021

DuReader_robust: A Chinese Dataset Towards Evaluating Robustness and Generalization of Machine Reading Comprehension in Real-World Applications

Machine reading comprehension (MRC) is a crucial task in natural language processing and has achieved remarkable advancements. However, most of the neural MRC models are still far from robust and fail to generalize well in real-world applications. In order to comprehensively verify the robustness and generalization of MRC models, we introduce a real-world Chinese dataset -- DuReader_robust. It is designed to evaluate the MRC models from three aspects: over-sensitivity, over-stability and generalization. Comparing to previous work, the instances in DuReader_robust are natural texts, rather than the altered unnatural texts. It presents the challenges when applying MRC models to real-world applications. The experimental results show that MRC models do not perform well on the challenge test set. Moreover, we analyze the behavior of existing models on the challenge test set, which may provide suggestions for future model development. The dataset and codes are publicly available at https://github.com/baidu/DuReader.

cs.CL↗

arXiv2021

Calculate Center-of-Inertia Frequency and System RoCoF Using PMU Data

The power system frequency is important for the system overall stability. However, there does not exist a single measurement point of the system frequency due to the distributed nature of the system inertia and the small inconsistency of different generator rotor electrical speeds in one synchronized system. This paper proposed a new approach to calculate the system center-of-inertia (COI) frequency and the rate-of-change-of-frequency (RoCoF) more accurately using PMU data at multiple locations. The COI frequency and the RoCoF value were further used to assist fast estimation of the imbalance MW amount of a frequency event. Test results using actual measurements in the U.S. Eastern Interconnection system validated the effectiveness of the proposed method.

eess.SY↗

arXiv2021

SuperDTI: Ultrafast diffusion tensor imaging and fiber tractography with deep learning

Purpose: To propose a deep learning-based reconstruction framework for ultrafast and robust diffusion tensor imaging and fiber tractography. Methods: We propose SuperDTI to learn the nonlinear relationship between diffusion-weighted images (DWIs) and the corresponding tensor-derived quantitative maps as well as the fiber tractography. Super DTI bypasses the tensor fitting procedure, which is well known to be highly susceptible to noise and motion in DWIs. The network is trained and tested using datasets from Human Connectome Project and patients with ischemic stroke. SuperDTI is compared against the state-of-the-art methods for diffusion map reconstruction and fiber tracking. Results: Using training and testing data both from the same protocol and scanner, SuperDTI is shown to generate fractional anisotropy and mean diffusivity maps, as well as fiber tractography, from as few as six raw DWIs. The method achieves a quantification error of less than 5% in all regions of interest in white matter and gray matter structures. We also demonstrate that the trained neural network is robust to noise and motion in the testing data, and the network trained using healthy volunteer data can be directly applied to stroke patient data without compromising the lesion detectability. Conclusion: This paper demonstrates the feasibility of superfast diffusion tensor imaging and fiber tractography using deep learning with as few as six DWIs directly, bypassing tensor fitting. Such a significant reduction in scan time may allow the inclusion of DTI into the clinical routine for many potential applications.

eess.IV↗

arXiv2021

Symbol-Level Precoding Design for Intelligent Reflecting Surface Assisted Multi-user MIMO Systems

Intelligent reflecting surface (IRS) has emerged as a promising solution to enhance wireless information transmissions by adaptively controlling prorogation environment. Recently, the brand-new concept of utilizing IRS to implement a passive transmitter attracts researchers' attention since it potentially realizes low-complexity and hardware-efficient transmitters of multiple-input single/multiple-output (MISO/MIMO) systems. In this paper we investigate the problem of precoder design for a low-resolution IRS-based transmitter to implement multi-user MISO/MIMO wireless communications. Particularly, the IRS modulates information symbols by varying the phases of its reflecting elements and transmits them to $K$ single-antenna or multi-antenna users. We first aim to design the symbol-level precoder for IRS to realize the modulation and minimize the maximum symbol-error-rate (SER) of single-antenna receivers. In order to tackle this NP-hard problem, we first relax the low-resolution phase-shift constraint and solve this problem by Riemannian conjugate gradient (RCG) algorithm. Then, the low-resolution symbol-level precoding vector is obtained by direct quantization. Considering the large quantization error for 1-bit resolution case, the branch-and-bound method is utilized to solve the 1-bit resolution symbol-level precoding vector. For multi-antenna receivers, we propose to iteratively design the symbol-level precoder and combiner by decomposing the original large-scale optimization problem into several sub-problems. Simulation results validate the effectiveness of our proposed algorithms.

eess.SP↗

arXiv2021

Channel Estimation for Practical IRS-Assisted OFDM Systems

Intelligent reflecting surface (IRS), composed of a large number of hardware-efficient passive elements, is deemed as a potential technique for future wireless communications since it can adaptively enhance the propagation environment. In order to effectively utilize IRS to achieve promising beamforming gains, the problem of channel state information (CSI) acquisition needs to be carefully considered. However, most recent works assume to employ an ideal IRS, i.e., each reflecting element has constant amplitude, variable phase shifts, as well as the same response for the signals with different frequencies, which will cause severe estimation error due to the mismatch between the ideal IRS and the practical one. In this paper, we study channel estimation in practical IRS-aided orthogonal frequency division multiplexing (OFDM) systems with discrete phase shifts. Different from the prior works which assume that IRS has an ideal reflection model, we perform channel estimation by considering amplitude-phase shift-frequency relationship for the response of practical IRS. Aiming at minimizing normalized-mean-square-error (NMSE) of the estimated channel, a novel IRS time-varying reflection pattern is designed by leveraging the alternating optimization (AO) algorithm for the case of using low-resolution phase shifters. Moreover, for the high-resolution IRS cases, we provide another practical reflection pattern scheme to further reduce the complexity. Simulation results demonstrate the necessity of considering practical IRS model for channel estimation and the effectiveness of our proposed channel estimation methods.

eess.SP↗

arXiv2021

Intelligent reflecting surface enhanced wideband MIMO-OFDM communications: From practical model to reflection optimization

Intelligent reflecting surface (IRS) is envisioned as a revolutionary technology for future wireless communication systems since it can intelligently change radio environment and integrate it into wireless communication optimization. However, most existing works adopted an ideal IRS reflection model, which is impractical and can cause significant performance degradation in realistic wideband systems. To address this issue, we first study the dual phase- and amplitude-squint effect of reflected signals and present a simplified practical IRS reflection model for wideband signals. Then, an IRS enhanced wideband multiuser multi-input single-output orthogonal frequency division multiplexing (MU-MISO-OFDM) system is investigated. We aim to jointly design the transmit beamformer and IRS reflection for the case of using both continuous and discrete phase shifters to maximize the average sum-rate over all subcarriers. By exploiting the relationship between sum-rate maximization and mean square error (MSE) minimization, the original problem is equivalently transformed into a multi-block/variable problem, which can be efficiently solved by the block coordinate descent (BCD) method. Complexity and convergence for both cases are analyzed or illustrated. Simulation results demonstrate that the proposed algorithm can offer significant average sum-rate enhancement compared to that achieved using the ideal IRS reflection model, which confirms the importance of the use of the practical model for the design of wideband systems.

eess.SP↗

arXiv2020

Learning Autocompletion from Real-World Datasets

Code completion is a popular software development tool integrated into all major IDEs. Many neural language models have achieved promising results in completion suggestion prediction on synthetic benchmarks. However, a recent study When Code Completion Fails: a Case Study on Real-World Completions demonstrates that these results may not translate to improvements in real-world performance. To combat this effect, we train models on real-world code completion examples and find that these models outperform models trained on committed source code and working version snapshots by 12.8% and 13.8% accuracy respectively. We observe this improvement across modeling technologies and show through A/B testing that it corresponds to a 6.2% increase in programmers' actual autocompletion usage. Furthermore, our study characterizes a large corpus of logged autocompletion usages to investigate why training on real-world examples leads to stronger models.

cs.SE↗

arXiv2020

A Study on Evaluation Standard for Automatic Crack Detection Regard the Random Fractal

A reasonable evaluation standard underlies construction of effective deep learning models. However, we find in experiments that the automatic crack detectors based on deep learning are obviously underestimated by the widely used mean Average Precision (mAP) standard. This paper presents a study on the evaluation standard. It is clarified that the random fractal of crack disables the mAP standard, because the strict box matching in mAP calculation is unreasonable for the fractal feature. As a solution, a fractal-available evaluation standard named CovEval is proposed to correct the underestimation in crack detection. In CovEval, a different matching process based on the idea of covering box matching is adopted for this issue. In detail, Cover Area rate (CAr) is designed as a covering overlap, and a multi-match strategy is employed to release the one-to-one matching restriction in mAP. Extended Recall (XR), Extended Precision (XP) and Extended F-score (Fext) are defined for scoring the crack detectors. In experiments using several common frameworks for object detection, models get much higher scores in crack detection according to CovEval, which matches better with the visual performance. Moreover, based on faster R-CNN framework, we present a case study to optimize a crack detector based on CovEval standard. Recall (XR) of our best model achieves an industrial-level at 95.8, which implies that with reasonable standard for evaluation, the methods for object detection are with great potential for automatic industrial inspection.

cs.CV↗

arXiv2020

Constraining the inner density slope of massive galaxy clusters

We determine the inner density profiles of massive galaxy clusters (M$_{200}$ > $5 \times 10^{14}$ M$_{\odot}$) in the Cluster-EAGLE (C-EAGLE) hydrodynamic simulations, and investigate whether the dark matter density profiles can be correctly estimated from a combination of mock stellar kinematical and gravitational lensing data. From fitting mock stellar kinematics and lensing data generated from the simulations, we find that the inner density slopes of both the total and the dark matter mass distributions can be inferred reasonably well. We compare the density slopes of C-EAGLE clusters with those derived by Newman et al. for 7 massive galaxy clusters in the local Universe. We find that the asymptotic best-fit inner slopes of "generalized" NFW (gNFW) profiles, $γ_{\rm gNFW}$, of the dark matter haloes of the C-EAGLE clusters are significantly steeper than those inferred by Newman et al. However, the mean mass-weighted dark matter density slopes of the simulated clusters are in good agreement with the Newman et al. estimates. We also find that the estimate of $γ_{\rm gNFW}$ is very sensitive to the constraints from weak lensing measurements in the outer parts of the cluster and a bias can lead to an underestimate of $γ_{\rm gNFW}$.

astro-ph.CO↗

arXiv2020

Practical Modeling and Beamforming for Intelligent Reflecting Surface Aided Wideband Systems

Intelligent reflecting surface (IRS) has emerged as a revolutionizing solution to enhance wireless communications by intelligently changing the propagation environment. Prior studies on IRS are based on an ideal reflection model with a constant amplitude and a variable phase shift. However, it is difficult and unrealistic to implement an IRS satisfying such ideal reflection model in practical applications. In this letter, we aim to investigate the phase-amplitude-frequency relationship of the reflected signals and propose a practical model of reflection coefficient for an IRS-aided wideband system. Then, based on this practical model, joint transmit power allocation of each subcarrier and IRS beamforming optimization are investigated for an IRS-aided wideband orthogonal frequency-division multiplexing (OFDM) system. Simulation results illustrate the importance of the practical model on the IRS designs and validate the effectiveness of our proposed model.

eess.SP↗

arXiv2020

Knowledge Federation: A Unified and Hierarchical Privacy-Preserving AI Framework

With strict protections and regulations of data privacy and security, conventional machine learning based on centralized datasets is confronted with significant challenges, making artificial intelligence (AI) impractical in many mission-critical and data-sensitive scenarios, such as finance, government, and health. In the meantime, tremendous datasets are scattered in isolated silos in various industries, organizations, different units of an organization, or different branches of an international organization. These valuable data resources are well underused. To advance AI theories and applications, we propose a comprehensive framework (called Knowledge Federation - KF) to address these challenges by enabling AI while preserving data privacy and ownership. Beyond the concepts of federated learning and secure multi-party computation, KF consists of four levels of federation: (1) information level, low-level statistics and computation of data, meeting the requirements of simple queries, searching and simplistic operators; (2) model level, supporting training, learning, and inference; (3) cognition level, enabling abstract feature representation at various levels of abstractions and contexts; (4) knowledge level, fusing knowledge discovery, representation, and reasoning. We further clarify the relationship and differentiation between knowledge federation and other related research areas. We have developed a reference implementation of KF, called iBond Platform, to offer a production-quality KF platform to enable industrial applications in finance, insurance et al. The iBond platform will also help establish the KF community and a comprehensive ecosystem and usher in a novel paradigm shift towards secure, privacy-preserving and responsible AI. As far as we know, knowledge federation is the first hierarchical and unified framework for secure multi-party computing and learning.

cs.CR↗

arXiv2020

Arbitrary-sized Image Training and Residual Kernel Learning: Towards Image Fraud Identification

Preserving original noise residuals in images are critical to image fraud identification. Since the resizing operation during deep learning will damage the microstructures of image noise residuals, we propose a framework for directly training images of original input scales without resizing. Our arbitrary-sized image training method mainly depends on the pseudo-batch gradient descent (PBGD), which bridges the gap between the input batch and the update batch to assure that model updates can normally run for arbitrary-sized images. In addition, a 3-phase alternate training strategy is designed to learn optimal residual kernels for image fraud identification. With the learnt residual kernels and PBGD, the proposed framework achieved the state-of-the-art results in image fraud identification, especially for images with small tampered regions or unseen images with different tampering distributions.

cs.CV↗

arXiv2019

When Deep Learning Met Code Search

There have been multiple recent proposals on using deep neural networks for code search using natural language. Common across these proposals is the idea of $\mathit{embedding}$ code and natural language queries, into real vectors and then using vector distance to approximate semantic correlation between code and the query. Multiple approaches exist for learning these embeddings, including $\mathit{unsupervised}$ techniques, which rely only on a corpus of code examples, and $\mathit{supervised}$ techniques, which use an $\mathit{aligned}$ corpus of paired code and natural language descriptions. The goal of this supervision is to produce embeddings that are more similar for a query and the corresponding desired code snippet. Clearly, there are choices in whether to use supervised techniques at all, and if one does, what sort of network and training to use for supervision. This paper is the first to evaluate these choices systematically. To this end, we assembled implementations of state-of-the-art techniques to run on a common platform, training and evaluation corpora. To explore the design space in network complexity, we also introduced a new design point that is a $\mathit{minimal}$ supervision extension to an existing unsupervised technique. Our evaluation shows that: 1. adding supervision to an existing unsupervised technique can improve performance, though not necessarily by much; 2. simple networks for supervision can be more effective that more sophisticated sequence-based networks for code search; 3. while it is common to use docstrings to carry out supervision, there is a sizeable gap between the effectiveness of docstrings and a more query-appropriate supervision corpus. The evaluation dataset is now available at arXiv:1908.09804

cs.SE↗

arXiv2019

Neural Code Search Evaluation Dataset

There has been an increase of interest in code search using natural language. Assessing the performance of such code search models can be difficult without a readily available evaluation suite. In this paper, we present an evaluation dataset consisting of natural language query and code snippet pairs, with the hope that future work in this area can use this dataset as a common benchmark. We also provide the results of two code search models ([1] and [6]) from recent work. The evaluation dataset is available at https://github.com/facebookresearch/Neural-Code-Search-Evaluation-Dataset

cs.SE↗

arXiv2019

IRS-Enhanced Wideband MU-MISO-OFDM Communication Systems

Intelligent reflecting surface (IRS) is considered as an enabling technology for future wireless communication systems since it can intelligently change the wireless environment to improve the communication performance. In this paper, an IRS-enhanced wideband multiuser multi-input single-output orthogonal frequency division multiplexing (MU-MISO-OFDM) system is investigated. We aim to jointly design the transmit beamformer and the reflection of IRS to maximize the average sum-rate over all subcarriers. With the aid of the relationship between sum-rate maximization and mean square error (MSE) minimization, an efficient joint beamformer and IRS design algorithm is developed. Simulation results illustrate that the proposed algorithm can offer significant average sum-rate enhancement, which confirms the effectiveness of the use of the IRS for wideband wireless communication systems.

eess.SP↗

arXiv2019

An End-to-End Encrypted Neural Network for Gradient Updates Transmission in Federated Learning

Federated learning is a distributed learning method to train a shared model by aggregating the locally-computed gradient updates. In federated learning, bandwidth and privacy are two main concerns of gradient updates transmission. This paper proposes an end-to-end encrypted neural network for gradient updates transmission. This network first encodes the input gradient updates to a lower-dimension space in each client, which significantly mitigates the pressure of data communication in federated learning. The encoded gradient updates are directly recovered as a whole, i.e. the aggregated gradient updates of the trained model, in the decoding layers of the network on the server. In this way, gradient updates encrypted in each client are not only prevented from interception during communication, but also unknown to the server. Based on the encrypted neural network, a novel federated learning framework is designed in real applications. Experimental results show that the proposed network can effectively achieve two goals, privacy protection and data compression, under a little sacrifice of the model accuracy in federated learning.

cs.LG↗

arXiv2019

Frequency Observations and Statistic Analysis of Worldwide Main Power Grids Using FNET/GridEye

With the increasing renewable energy sources, concerns about how renewable energy sources impact frequency have risen. There are few reports regarding power frequency status in worldwide main power grids and what are differences of frequency status between power grids in mainland and island. FNET/GridEye, a wide-area measurement system collecting frequency and phase angle data at the distribution level, provides an opportunity to observe and study the power frequency in different power grids over the world. In this paper, 13 different power grids, spreading at different mainland and islands over the world, are observed and compared. A more detail statistical analysis was conducted for typical power grids in three different places, e.g., U.S Eastern Interconnection (EI), Egypt, and Japan. The probability functions of frequency based on the measured data are calculated. The distributions of frequency in different power grids fall into two categories, e.g., single-peak distribution and multi-peak distribution. Furthermore, a meaningful insight that the single-peak distributions of the frequency almost follow the normal distribution is found. The frequency observations and statistic analysis of worldwide main power grids using FNET/GridEye could help the power system operators understand the frequency statistical characteristic more deeply.

eess.SY↗

arXiv2019

Photometric and Kinematic Misalignments and Their Evolution Among Fast and Slow Rotators in the Illustris Simulation

We use the Illustris simulation to study the distributions of ellipticities and kinematic misalignments of galactic projections, as well as the intrinsic shapes and rotation of the simulated galaxies. Our results for the projections of galaxies display clear trends of an overall increase of kinematic misalignment and a slight decrease of ellipticity for fast rotators with increasing masses, while revealing no clear distinction between slow rotators of different mass. It is also found that the number of very slow rotators with large ellipticities is much larger than found in observations. The intrinsic properties of the galaxies are then analysed. The results for the intrinsic shapes of the galaxies are mostly consistent with previous results inferred from observational data. The distributions of intrinsic misalignment of the galaxies suggest that some of the galaxies produced by Illustris have significant rotation around their medium axes. Further analysis reveals that most of these galaxies display signs of non-equilibrium. We then study the evolution of the intrinsic misalignments and shapes of three specific Illustris galaxies, which we consider as typical ones, along the main progenitor line of their merger trees, revealing how mergers influence the intrinsic shapes and kinematics: the spin axis in general stays close to the shortest axis, and tends to quickly relax to such an equilibrium state within a few dynamical times of the galaxy after major perturbations; triaxiality and intrinsic flatness in general decrease with time, however, sometimes increases occur that are clearly seen to correlate with major merger events.

astro-ph.GA↗