A5100664162- Topic Modeling
- Natural Language Processing Techniques
- Nuclear Physics and Applications
- Advanced Image Processing Techniques
- Web Data Mining and Analysis
- Anomaly Detection Techniques and Applications
- Image and Signal Denoising Methods
- Adversarial Robustness in Machine Learning
- Radiation Detection and Scintillator Technologies
- Image Enhancement Techniques
- Advanced Image and Video Retrieval Techniques
- Advanced Vision and Imaging
- Radiation Therapy and Dosimetry
- Nuclear reactor physics and engineering
- Reinforcement Learning in Robotics
- Text and Document Classification Technologies
- Domain Adaptation and Few-Shot Learning
- Service-Oriented Architecture and Web Services
- Face and Expression Recognition
- Optical Systems and Laser Technology
- Multimodal Machine Learning Applications
- Advanced Text Analysis Techniques
- Sentiment Analysis and Opinion Mining
- Face recognition and analysis
- Particle Detector Development and Performance
University of Science and Technology of China
2008-2024
Nanjing University of Science and Technology
2024
Liupanshui Normal University
2024
National University of Singapore
2023
Yale University
2017-2021
East China University of Technology
2021
Tibetan Traditional Medical College
2021
Salesforce (United States)
2019
Xi'an Jiaotong University
2019
Tsinghua University
2019
Graph Convolutional Networks (GCNs) and their variants have experienced significant attention become the de facto methods for learning graph representations. GCNs derive inspiration primarily from recent deep approaches, as a result, may inherit unnecessary complexity redundant computation. In this paper, we reduce excess through successively removing nonlinearities collapsing weight matrices between consecutive layers. We theoretically analyze resulting linear model show that it corresponds...
Tao Yu, Rui Zhang, Kai Yang, Michihiro Yasunaga, Dongxu Wang, Zifan Li, James Ma, Irene Qingning Yao, Shanelle Roman, Zilin Dragomir Radev. Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing. 2018.
Tao Yu, Zifan Li, Zilin Zhang, Rui Dragomir Radev. Proceedings of the 2018 Conference North American Chapter Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers). 2018.
Most existing studies in text-to-SQL tasks do not require generating complex SQL queries with multiple clauses or sub-queries, and generalizing to new, unseen databases. In this paper we propose SyntaxSQLNet, a syntax tree network address the cross-domain generation task. SyntaxSQLNet employs specific tree-based decoder path history table-aware column attention encoders. We evaluate on new large-scale corpus containing databases tables nested queries. use database split setting where test...
Microbial fuel cells (MFCs) are devices that use bacteria as the catalysts to oxidize organic and inorganic matter generate current whereas microbial electrolysis (MECs) a reactor for biohydrogen production by combining MFC electrolysis. In an MEC, external voltage must be applied overcome thermodynamic barrier. Here we report MEC-MFC-coupled system from acetate, in which hydrogen was produced MEC extra power supplied MFC. this coupled system, acetate without electric supply. At 10 mM of...
Rui Zhang, Tao Yu, Heyang Er, Sungrok Shim, Eric Xue, Xi Victoria Lin, Tianze Shi, Caiming Xiong, Richard Socher, Dragomir Radev. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and 9th International Joint (EMNLP-IJCNLP). 2019.
A good distortion representation is crucial for the success of deep blind image quality assessment (BIQA). However, most previous methods do not effectively model relationship between distortions or distribution samples with same type but different levels. In this work, we start from analysis perceptual and distortion-related factors, such as types Then, propose a Distortion Graph Representation (DGR) learning framework IQA, named GraphIQA, in which each represented graph, <italic...
Large-scale vision-language models (VLMs) pre-trained on billion-level data have learned general visual representations and broad concepts. In principle, the welllearned knowledge structure of VLMs should be inherited appropriately when being transferred to downstream tasks with limited data. However, most existing efficient transfer learning (ETL) approaches for either damage or are excessively biased towards prior knowledge, e.g., prompt tuning (PT) discards text-based classifier builds a...
Back-streaming neutrons through the incoming proton channel at spallation target station of China Spallation Neutron Source (CSNS) has been exploited as a white neutron beam line (so-called Back-n), and number spectrometers for nuclear data measurements have planned. With thick tungsten modest moderation by cooling water slices, is very intense which in order 5.0×106 n/cm2/s 80 m from an excellent energy spectrum spanning 1 eV to 100 MeV. In addition, time structure primary under different...
Natural images are virtually surrounded by low-density misclassified regions that can be efficiently discovered gradient-guided search --- enabling the generation of adversarial images. While many techniques for detecting these attacks have been proposed, they easily bypassed when adversary has full knowledge detection mechanism and adapts attack strategy accordingly. In this paper, we adopt a novel perspective regard omnipresence perturbations as strength rather than weakness. We postulate...
We present GraPPa, an effective pre-training approach for table semantic parsing that learns a compositional inductive bias in the joint representations of textual and tabular data. construct synthetic question-SQL pairs over high-quality tables via synchronous context-free grammar (SCFG) induced from existing text-to-SQL datasets. pre-train our model on data using novel text-schema linking objective predicts syntactic role field SQL each pair. To maintain model's ability to represent...
Traditional single image super-resolution (SISR) methods that focus on solving and uniform degradation (i.e., bicubic down-sampling), typically suffer from poor performance when applied into real-world low-resolution (LR) images due to the complicated realistic degradations. The key this more challenging real (RealSR) problem lies in learning feature representations are both informative content-aware. In paper, we propose a Omni-frequency Region-adaptive Network (OR-Net) address challenges,...
Recently, learning-based algorithms for image inpainting achieve remarkable progress dealing with squared or irregular holes. However, they fail to generate plausible textures inside damaged area because there lacks surrounding information. A progressive approach would be advantageous eliminating central blurriness, i.e., restoring well and then updating masks. In this paper, we propose full-resolution residual network (FRRN) fill holes, which is proved effective inpainting. We show that...
Data augmentation (DA) plays a critical role in improving the generalization of deep learning models. Recent works on automatically searching for DA policies from data have achieved great success. However, existing automated methods generally perform search at image level, which limits exploration diversity local regions. In this paper, we propose more fine-grained approach, dubbed Patch AutoAugment, to divide an into grid patches and joint optimal patches. We formulate it as multi-agent...
English text has a clear and compact subject structure, which makes it easy to find dependency relationships between words. However, Chinese often conveys information using situational settings, results in loose sentence structures, even most comments experimental summary texts lack subjects. This challenging determine the relationship words text, especially aspect-level sentiment recognition. To solve this problem faced by field of recognition, dual attention network for recognition is...