Hao Yu

ORCID: 0000-0003-0709-8947
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Natural Language Processing Techniques
  • Topic Modeling
  • Web Data Mining and Analysis
  • Advanced Text Analysis Techniques
  • Semantic Web and Ontologies
  • Sentiment Analysis and Opinion Mining
  • Oceanographic and Atmospheric Processes
  • Climate variability and models
  • Text and Document Classification Technologies
  • Methane Hydrates and Related Phenomena
  • Meteorological Phenomena and Simulations
  • Speech and dialogue systems
  • Algorithms and Data Compression
  • Distributed and Parallel Computing Systems
  • Biomedical Text Mining and Ontologies
  • Advanced Computational Techniques and Applications
  • Ocean Waves and Remote Sensing
  • Privacy-Preserving Technologies in Data
  • High voltage insulation and dielectric phenomena
  • Spam and Phishing Detection
  • Energy Load and Power Forecasting
  • Geophysics and Gravity Measurements
  • Data Quality and Management
  • Image Retrieval and Classification Techniques
  • Computational Geometry and Mesh Generation

Shanghai Jiao Tong University
2005-2025

Tsinghua University
2007-2024

Beijing University of Posts and Telecommunications
2011-2024

Nanjing University of Aeronautics and Astronautics
2023-2024

Smile Train
2023

Chongqing University of Posts and Telecommunications
2023

PLA Information Engineering University
2023

Bozhou People's Hospital
2023

Anhui University of Traditional Chinese Medicine
2023

University of Electronic Science and Technology of China
2022

We study the problem of extracting correspondences between a pair point clouds for registration. For correspondence retrieval, existing works benefit from matching sparse keypoints detected dense points but usually struggle to guarantee their repeatability. To address this issue, we present CoFiNet - Coarse-to-Fine Network which extracts hierarchical coarse fine without keypoint detection. On scale and guided by weighting scheme, our model firstly learns match down-sampled nodes whose...

10.48550/arxiv.2110.14076 preprint EN other-oa arXiv (Cornell University) 2021-01-01

Background Smartphone-based online education gained popularity during and after the COVID-19 pandemic. Although recent studies have highlighted association between problematic smartphone use (PSU) mental health symptoms, potential role of learning in this relationship remains unclear. This study aimed t o analyze relationships higher modes, PSU, related psychological symptoms university students. Methods A total 1,629 Chinese students from five provinces completed a web-based questionnaire...

10.3389/fpubh.2022.881074 article EN cc-by Frontiers in Public Health 2022-05-04

Heralded single photon source (HSPS) is an important way in generating genuine photon, having advantages of experimental simplicity and versatility. However, HSPS intrinsically suffers from the trade-off between heralded rate purity. To overcome this, one can apply multiplexing technology different degrees freedom to enhance performance HSPS. Here, by employing spectral active feed-forward manipulating, we demonstrate a at 1.5 {\mu}m telecom-band. Our results show that effectively erases...

10.1364/prj.450731 article EN Photonics Research 2022-04-04

Compared to traditional metal cable, high-temperature superconductor (HTS) cable is a promising candidate for the energy transmission in space solar power stations due its great advantage high density and efficiency. These cables can reduce losses simplify conventional by eliminating need voltage conversion equipment, thus reducing launch weight costs of spacecraft. This paper analyzes feasibility superconducting spacecraft transfer. Addressing operating conditions vacuum cryogenic...

10.1109/tasc.2024.3525446 article EN IEEE Transactions on Applied Superconductivity 2025-01-01

This paper proposes a hybrid Chinese named entity recognition model based on multiple features. It differentiates from most of the previous approaches mainly as follows. Firstly, proposed Hybrid Model integrates coarse particle feature (POS Model) with fine (Word Model), so that it can overcome disadvantages each other. Secondly, in order to reduce searching space and improve efficiency, we introduce heuristic human knowledge into statistical model, which could increase performance NER...

10.3115/1220575.1220629 article EN 2005-01-01

Abstract Tropical cyclones (TCs), especially landfalling intense storms often pose serious threats to life and property in coastal areas. Although TC track forecast skill has been improved the past decades, progress of advancing intensity lags that forecast. One possible limiting factor is absence ocean surface waves systems. To better represent interaction underlying ocean, a regional atmosphere‐ocean‐wave coupled model employed this study. Twenty‐one TCs whole year 2013 are retrospectively...

10.1029/2022jc019015 article EN cc-by-nc Journal of Geophysical Research Oceans 2022-11-01

Federated Learning (FL) offers collaborative model training without data sharing but is vulnerable to backdoor attacks, where poisoned weights lead compromised system integrity. Existing countermeasures, primarily based on anomaly detection, are prone erroneous rejections of normal while accepting ones, largely due shortcomings in quantifying similarities among client models. Furthermore, other defenses demonstrate effectiveness only when dealing with a limited number malicious clients,...

10.48550/arxiv.2306.04984 preprint EN other-oa arXiv (Cornell University) 2023-01-01

This paper proposes a new discriminative training method, called minimum sample risk (MSR), of estimating parameters language models for text input. While most existing methods use loss function that can be optimized easily but approaches only approximately to the objective error rate, MSR minimizes directly using heuristic procedure. Evaluations on task Japanese input show handle large number features and samples; it significantly outperforms regular trigram model trained maximum likelihood...

10.3115/1220575.1220602 article EN 2005-01-01

The recently proposed weak form quadrature element method (QEM) is applied to the buckling analysis of planar frameworks. This starts with approximation integrands in description (variational formulation) a problem. Neither nodes nor number fixed, being adjustable according convergence need. Examples are presented and comparison results finite made demonstrate effectiveness computational efficiency QEM. It shown that QEM suitable for frameworks either varying or constant cross sections.

10.1142/s0219455411004130 article EN International Journal of Structural Stability and Dynamics 2011-03-30

Transient voltages in the power grid are key for fault analysis of a grid, optimized insulation design, and standardization high-voltage testing method. The traditional measuring equipment, based on electrical engineering, normally has limited bandwidth response speed, which also featured by huge size heavy weight. In this paper, an integrated optical electric-field sensor Pockels effect was developed applied to measure transient conductors non-contact mode. system speed faster than 6 ns...

10.3390/app9091951 article EN cc-by Applied Sciences 2019-05-13

Locating faults in transmission lines is important for the safety and continuous operation of power systems. The existing fault location methods based on traveling waves mostly use current transformers to obtain transient signals, which requires a direct connection with live therefore brings inconvenience installation, debugging, maintenance. In this article, we propose flexible noncontact approach locating broadband electro-optic field sensors, where system can be easily installed,...

10.1109/temc.2021.3108938 article EN IEEE Transactions on Electromagnetic Compatibility 2021-09-23

Abstract. This article describes the implementation of a coupling between global forecast model (CFSv2.0) and wave (WW3) investigates effects ocean surface waves on air–sea interface in new framework. Several major wave-related processes, including Langmuir mixing, Stokes–Coriolis force with entrainment, fluxes modified by Stokes drift, momentum roughness length, are evaluated two groups 56 d experiments, one for boreal winter other summer. Comparisons made against situ buoys, satellite...

10.5194/gmd-15-2345-2022 article EN cc-by Geoscientific model development 2022-03-18

For web content extraction task, researchers have proposed many different methods, such as wrapper-based method, DOM tree rule-based machine learning-based method and so on. To some extent, all these methods ignore the layout information of webpage, although spatial visual cues often plays a very important role in process locating main webpage when browsing. As consequence, throw part away extracting from webpage. In this paper, we present which combines analysis with rule-base it can make...

10.1109/itcs.2010.16 article EN 2010-07-01

Hierarchical phrase-based statistical machine translation systems often suffer from a huge rule table. This paper proposed basic and efficient method for table reduction, discarding monotone composed rules. These rules are redundant because they may be monotonically recreated by minimal Experiments show that the is reduced 57%~71% without worsening quality.

10.1145/1667780.1667786 article EN 2009-12-03

In this paper, we investigate how to identify domain-sensitive opinion leaders in online review communities, and present a model rank leaders. To evaluate the effectiveness of proposed model, conduct preliminary experiments on real-world dataset from Amazon.com. Experimental results indicate that is effective identifying

10.1145/2487788.2487882 article EN 2013-05-13

Abstract. The Arctic regional coupled sea-ice–ocean–atmosphere model (ArcIOAM) has been developed to provide reliable sea ice prediction on seasonal timescales. description and implementation of ArcIOAM its preliminary results for the year 2012 are presented in this paper. In configuration, Community Coupler 2 (C-Coupler2) is used couple sea-ice–oceanic configuration MITgcm (Massachusetts Institute Technology general circulation model) with atmospheric Polar WRF (Weather Research...

10.5194/gmd-14-1101-2021 article EN cc-by Geoscientific model development 2021-02-24

Using abundant Web resources to mine Chinese term translations can be applied in many fields such as reading/writing assistant, machine translation and cross-language information retrieval. In mining English of terms, how obtain effective pages evaluate candidates are two challenging issues. this paper, the approach based on semantic prediction is first proposed pages. The method predicts possible meanings according each constituent unit term, expands these items using semantically relevant...

10.3115/1273073.1273099 article EN 2006-01-01

Recently, the number of freely available online reviews is increasing in a high speed. More and more aspect based opinion mining technique has been employed to find out customers' opinions. In this paper, we only focus on categorize product features that customers have commented on. An unsupervised twice-clustering categorization method proposed. Opinion words context are chosen represent interrelationship among instead full information. The cluster result active used as constraints improve...

10.1109/wism.2010.71 article EN 2010-10-01

Introduction: Ginger (Zingiber officinale Roce.) is a widely consumed food item and prominent traditional Chinese medicinal herb. The intrinsic quality of ginger may differ due to variations in its origin processing techniques. To evaluate the ginger, straightforward efficient discriminatory approach has been devised, utilizing 6-gingerol, 8-gingerol, 10-gingerol as benchmarks. Methods: In order categorize samples according their cultivated origins with different longitude latitude...

10.3389/fchem.2023.1296712 article EN cc-by Frontiers in Chemistry 2023-11-10
Coming Soon ...