A5101922055- Topic Modeling
- Direction-of-Arrival Estimation Techniques
- Radar Systems and Signal Processing
- Natural Language Processing Techniques
- Speech and Audio Processing
- Advanced SAR Imaging Techniques
- Antenna Design and Optimization
- Speech Recognition and Synthesis
- Multimodal Machine Learning Applications
- Indoor and Outdoor Localization Technologies
- Underwater Acoustics Research
- Intelligent Tutoring Systems and Adaptive Learning
- Text Readability and Simplification
- Microwave Imaging and Scattering Analysis
- Advanced Text Analysis Techniques
- Domain Adaptation and Few-Shot Learning
- Advanced Adaptive Filtering Techniques
- Cognitive Science and Mapping
- Advanced Measurement and Detection Methods
- Infrared Target Detection Methodologies
- Target Tracking and Data Fusion in Sensor Networks
- Speech and dialogue systems
- Phonetics and Phonology Research
- Advanced Graph Neural Networks
- Text and Document Classification Technologies
Air Force Engineering University
2007-2023
University of Science and Technology of China
2005-2006
University of Calgary
2002
Cloze-style reading comprehension is a representative problem in mining relationship between document and query. In this paper, we present simple but novel model called attention-over-attention reader for better solving cloze-style task. The proposed aims to place another attention mechanism over the document-level induces “attended attention” final answer predictions. One advantage of our that it simpler than related works while giving excellent performance. addition primary model, also...
This paper addresses the issue of two-dimensional (2-D) direction arrival (DOA) estimation with coprime planar arrays (CPPAs) via sparse representation. Our work differs from partial spectral search approach [25], which suppresses phase ambiguity by searching common peaks two subarrays. We focus on property CPPA, where array extension model sum-difference coarray (SDCA) is derived for larger degrees freedom (DOFs). Besides, to optimize selection regularization parameter, we also construct a...
Recommending suitable exercises to students in an online education system is highly useful. Existing approaches usually rely on machine learning techniques mine large amounts of student interaction log data accumulated the systems select most for each student. Generally, they mainly aim optimize a single objective, i.e., recommending non-mastered address immediate weakness students. While this reasonable there exist more beneficial multiple objectives long-term process that need be addressed...
For the problem of joint angle and range estimation with frequency diverse array (FDA), MIMO radar, staggered increment is proposed to expand ambiguity algorithm ESPRIT MUSIC reduce computational complexity. The uniformly weighted beampattern FDA a SINC-like function. Therefore, grating lobe exists. In this paper, increase distance adjacent lobes. firstly estimates by using algorithm. Then we get one-dimensional search above estimated angle. simulation results section, it indicated in that...
Discourse modes play an important role in writing composition and evaluation. This paper presents a study on the manual automatic identification of narration,exposition, description, argument emotion expressing sentences narrative essays. We annotate corpus to characteristics discourse describe neural sequence labeling model for identification. Evaluation results show that can be identified automatically with average F1-score 0.7. further demonstrate used as features improve essay scoring...
Bo Chen, Xiaotao Gu, Yufeng Hu, Siliang Tang, Guoping Yueting Zhuang, Xiang Ren. Proceedings of the 2019 Conference North American Chapter Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers). 2019.
Owing to the continuous efforts by Chinese NLP community, more and machine reading comprehension datasets become available. To add diversity in this area, paper, we propose a new task called Sentence Cloze-style Machine Reading Comprehension (SC-MRC). The proposed aims fill right candidate sentence into passage that has several blanks. We built dataset CMRC 2019 evaluate difficulty of SC-MRC task. Moreover, difficulties, also made fake candidates are similar correct ones, which requires...
Abstract This letter proposes an efficient tensor model for Direction of Departure (DOD) and Arrival (DOA) estimation based on bistatic nested Multiple‐Input Multiple‐Output (MIMO) radar. Tensor offers a robust solution in characterizing multi‐dimensional data structures, ensuring integrity, has demonstrated remarkable effectiveness the realm parameter estimation. However, complexity traditional methods constructing tensors is major obstacle to practical applications. In response, this...
In this paper, we study the problem of coherent source localisation with multi-input multi-output (MIMO) radar via a smoothing matrix set (SMS) scheme. We first perform transmission diversity to form some sub-vectors (TSVs) and then reception for each TSV. That is, TSV consists (RSVs). By rearranging elements covariance sub-matrix along vertical direction, both forward-only SMS (FO-SMS) forward backward (FB-SMS) methods are derived using these RSVs. Compared existing methods, proposed scheme...
Machine Reading Comprehension (MRC) has become enormously popular recently and attracted a lot of attention. However, existing reading comprehension datasets are mostly in English. To add diversity datasets, this paper we propose new Chinese dataset for accelerating related research the community. The proposed contains two different types: cloze-style user query comprehension, associated with large-scale training data as well human-annotated validation hidden test set. Along dataset, also...
An optimized detection model based on weighted entropy for multiple input output (MIMO) radar in multipath environment is presented. After defining the distance difference (MDD), received signal with four paths built systematically. Both variance and correlation coefficient of scattering MDD are analyzed, which indicates that variable can decrease performance by reducing echo power. By making use likelihood ratio test (LRT), a new method introduced to positive power suppress negative power,...
This paper addresses the problem of two-dimensional direction-of-arrival (2D DOA) estimation low-elevation or non-low-elevation targets using L-shaped uniform and sparse arrays by analyzing signal models’ features their mapping to 2D DOA. proposes a DOA algorithm based on dilated convolutional network model, which consists two components: autoencoder neural network. If there are at low elevation, suppresses multipath outputs new covariance matrix as input directly perform in absence target....
Owing to the availability of various large-scale Machine Reading Comprehension ( MRC ) datasets, building an effective model extract passage spans for question answering has been well studied in previous works. However, reality, there are some questions that cannot be answered through information, which brings more challenges this task. In article, we propose Interactive Gated Decoder IG ), focuses on modeling interactions between answer span prediction and no-answer with a gating mechanism....
Due to the discontinuity of ocean waves and mountains, there are often multipath propagation effects obvious pulse characteristics in low-altitude detection. If conventional direction arrival (DOA) estimation method is directly used for finding, it will lead a large error. In view serious misalignment DOA signals under background impulse noise, based on spatial difference modified projection subspace algorithm proposed this paper. Firstly, covariance matrix received data vector eliminate...
Grammatical error diagnosis is an important task in natural language processing. This paper introduces our system at NLPTEA-2020 Task: Chinese Error Diagnosis (CGED). CGED aims to diagnose four types of grammatical errors which are missing words (M), redundant (R), bad word selection (S) and disordered (W). Our built on the model multi-layer bidirectional transformer encoder ResNet integrated into improve performance. We also explore two ensemble strategies including weighted averaging...
Retrieving information from correlative paragraphs or documents to answer open-domain multi-hop questions is very challenging. To deal with this challenge, most of the existing works consider as nodes in a graph and propose graph-based methods retrieve them. However, paper, we point out intrinsic defect such methods. Instead, new architecture that models sequential data considers retrieval kind sequence labeling task. Specifically, design rewritable external memory model dependency among...
In this paper, we initiated a method to estimate the direction of arrival (DOA) far-field, narrowband, and incoherent targets using coprime array. First, proposed array structure analysed distribution difference coarray (DCA). The degrees freedom (DOF) became clearer by referring DCA conception. However, previous algorithm only uses continuous virtual array, which causes elements in repeated position being abandoned. Therefore, paper analyses based on conception averages receiving signal...