Kaijie Zhou
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Kaijie Zhou has a total of 33 co-authors. Here are the top co-authors by number of shared publications:
- Yiying Yang
Ping An Life Insurance Company of China, Ltd Shenzhen China
3 shared publications - Haiqin Yang
International Digital Economy Academy (IDEA)
2 shared publications - Hao Peng
Beihang University, Beijing, China
2 shared publications Jianping Shen
Ping An Life Insurance of China, Ltd.,Shenzhen,China
2 shared publications- Jianping Shen
Ping An Life Insurance Company of China, Ltd Shenzhen China
2 shared publications - Xi Yin
Michigan State University
2 shared publications
Latest Publications
- 31 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 12 (20) , 42728-42740
Lightweight and Efficient Hybrid Network for UAV Identification Using Radio Frequency Fingerprinting
The widespread use of unmanned aerial vehicles brings both convenience and potential security risks, posing significant challenges for accurate UAV identification. To address the limitations of existing deep learning-based radio frequency fingerprinting methods in terms of computational complexity and model adaptability, this paper proposes an innovative hybrid model that combines Convolutional Neural Networks and Transformers to exploit both local and global features of RF signals fully. Our model consists of a Local Block and a Global Block. The Local Block employs Partial Convolution for feature extraction, PointWise Convolution to enhance feature representation, and the Squeeze-and-Excitation module to adaptively emphasize critical features, thereby improving local feature expressiveness. The Global Block comprises Unfold, Super Token Transformer Block, and Fold, which together enable effective modeling of global dependencies through signal unfolding, spatiotemporal transformations, and reconstruction. Experimental results show that under signal-to-noise ratio (SNR) conditions ranging from –5 dB to 20 dB, our method achieves an average recognition accuracy of 96.89%, with powerful performance under low SNR conditions. Furthermore, experiments demonstrate the model’s robustness with limited data. Even with only 1,000 training samples, the model maintains an accuracy of 92.14%. When incorporating Mixup data augmentation, high classification performance is sustained with just 500 training samples. These results highlight the method’s strong adaptability and practical potential in complex environments. Our codes and models are available at https://github.com/Zhoukaijie-hy/hybrid-model.
- 31 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 12 (20) , 42728-42740
Lightweight and Efficient Hybrid Network for UAV Identification Using Radio Frequency Fingerprinting
The widespread use of unmanned aerial vehicles brings both convenience and potential security risks, posing significant challenges for accurate UAV identification. To address the limitations of existing deep learning-based radio frequency fingerprinting methods in terms of computational complexity and model adaptability, this paper proposes an innovative hybrid model that combines Convolutional Neural Networks and Transformers to exploit both local and global features of RF signals fully. Our model consists of a Local Block and a Global Block. The Local Block employs Partial Convolution for feature extraction, PointWise Convolution to enhance feature representation, and the Squeeze-and-Excitation module to adaptively emphasize critical features, thereby improving local feature expressiveness. The Global Block comprises Unfold, Super Token Transformer Block, and Fold, which together enable effective modeling of global dependencies through signal unfolding, spatiotemporal transformations, and reconstruction. Experimental results show that under signal-to-noise ratio (SNR) conditions ranging from –5 dB to 20 dB, our method achieves an average recognition accuracy of 96.89%, with powerful performance under low SNR conditions. Furthermore, experiments demonstrate the model’s robustness with limited data. Even with only 1,000 training samples, the model maintains an accuracy of 92.14%. When incorporating Mixup data augmentation, high classification performance is sustained with just 500 training samples. These results highlight the method’s strong adaptability and practical potential in complex environments. Our codes and models are available at https://github.com/Zhoukaijie-hy/hybrid-model.
- 28 May 2023
- Published by Springer Nature
- p. 482-494
Semi-Supervised Text Classification via Self-Paced Semantic-Level Contrast
- 28 May 2023
- Published by Springer Nature
- p. 482-494
Semi-Supervised Text Classification via Self-Paced Semantic-Level Contrast
- 18 July 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-8
Aspect-Based Sentiment Classification with Background Information and Syntactic Auxiliary Tasks
Aspect-based sentiment classification is the task of predicting the sentiment tendency of a text toward a given aspect. Existing works on this task mainly focus on aspect-relevant information. In contrast, we design a model (BAT) which could extract overall Background information as well as Aspect-relevant informaTion. To make the BAT model learn better semantic representation of the given text, we introduce two auxiliary tasks (dependency neighborhood prediction and part-of-speech tagging). These auxiliary tasks are used to train the model together with the main sentiment classification task. Experiments on three benchmark datasets demonstrate that our method is effective and the proposed model achieves substantial performance improvements over comparison models.
- 18 July 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-8
Aspect-Based Sentiment Classification with Background Information and Syntactic Auxiliary Tasks
Aspect-based sentiment classification is the task of predicting the sentiment tendency of a text toward a given aspect. Existing works on this task mainly focus on aspect-relevant information. In contrast, we design a model (BAT) which could extract overall Background information as well as Aspect-relevant informaTion. To make the BAT model learn better semantic representation of the given text, we introduce two auxiliary tasks (dependency neighborhood prediction and part-of-speech tagging). These auxiliary tasks are used to train the model together with the main sentiment classification task. Experiments on three benchmark datasets demonstrate that our method is effective and the proposed model achieves substantial performance improvements over comparison models.
- 6 April 2021
- Published by Springer Nature
- p. 174-190
KGSynNet: A Novel Entity Synonyms Discovery Framework with Knowledge Graph
- 6 April 2021
- Published by Springer Nature
- p. 174-190
KGSynNet: A Novel Entity Synonyms Discovery Framework with Knowledge Graph
- Open Access
KGSynNet: A Novel Entity Synonyms Discovery Framework with Knowledge Graph
Entity synonyms discovery is crucial for entity-leveraging applications. However, existing studies suffer from several critical issues: (1) the input mentions may be out-of-vocabulary (OOV) and may come from a different semantic space of the entities; (2) the connection between mentions and entities may be hidden and cannot be established by surface matching; and (3) some entities rarely appear due to the long-tail effect. To tackle these challenges, we facilitate knowledge graphs and propose a novel entity synonyms discovery framework, named \emph{KGSynNet}. Specifically, we pre-train subword embeddings for mentions and entities using a large-scale domain-specific corpus while learning the knowledge embeddings of entities via a joint TransC-TransE model. More importantly, to obtain a comprehensive representation of entities, we employ a specifically designed \emph{fusion gate} to adaptively absorb the entities' knowledge information into their semantic features. We conduct extensive experiments to demonstrate the effectiveness of our \emph{KGSynNet} in leveraging the knowledge graph. The experimental results show that the \emph{KGSynNet} improves the state-of-the-art methods by 14.7\% in terms of hits@3 in the offline evaluation and outperforms the BERT model by 8.3\% in the positive feedback rate of an online A/B test on the entity linking module of a question answering system.
Open AccessKGSynNet: A Novel Entity Synonyms Discovery Framework with Knowledge Graph
Entity synonyms discovery is crucial for entity-leveraging applications. However, existing studies suffer from several critical issues: (1) the input mentions may be out-of-vocabulary (OOV) and may come from a different semantic space of the entities; (2) the connection between mentions and entities may be hidden and cannot be established by surface matching; and (3) some entities rarely appear due to the long-tail effect. To tackle these challenges, we facilitate knowledge graphs and propose a novel entity synonyms discovery framework, named \emph{KGSynNet}. Specifically, we pre-train subword embeddings for mentions and entities using a large-scale domain-specific corpus while learning the knowledge embeddings of entities via a joint TransC-TransE model. More importantly, to obtain a comprehensive representation of entities, we employ a specifically designed \emph{fusion gate} to adaptively absorb the entities' knowledge information into their semantic features. We conduct extensive experiments to demonstrate the effectiveness of our \emph{KGSynNet} in leveraging the knowledge graph. The experimental results show that the \emph{KGSynNet} improves the state-of-the-art methods by 14.7\% in terms of hits@3 in the offline evaluation and outperforms the BERT model by 8.3\% in the positive feedback rate of an online A/B test on the entity linking module of a question answering system.
- Open Access
An End-to-End Dialogue State Tracking System with Machine Reading Comprehension and Wide & Deep Classification
This paper describes our approach in DSTC 8 Track 4: Schema-Guided Dialogue State Tracking. The goal of this task is to predict the intents and slots in each user turn to complete the dialogue state tracking (DST) based on the information provided by the task's schema. Different from traditional stage-wise DST, we propose an end-to-end DST system to avoid error accumulation between the dialogue turns. The DST system consists of a machine reading comprehension (MRC) model for non-categorical slots and a Wide & Deep model for categorical slots. As far as we know, this is the first time that MRC and Wide & Deep model are applied to DST problem in a fully end-to-end way. Experimental results show that our framework achieves an excellent performance on the test dataset including 50% zero-shot services with a joint goal accuracy of 0.8652 and a slot tagging F1-Score of 0.9835.
Open AccessAn End-to-End Dialogue State Tracking System with Machine Reading Comprehension and Wide & Deep Classification
This paper describes our approach in DSTC 8 Track 4: Schema-Guided Dialogue State Tracking. The goal of this task is to predict the intents and slots in each user turn to complete the dialogue state tracking (DST) based on the information provided by the task's schema. Different from traditional stage-wise DST, we propose an end-to-end DST system to avoid error accumulation between the dialogue turns. The DST system consists of a machine reading comprehension (MRC) model for non-categorical slots and a Wide & Deep model for categorical slots. As far as we know, this is the first time that MRC and Wide & Deep model are applied to DST problem in a fully end-to-end way. Experimental results show that our framework achieves an excellent performance on the test dataset including 50% zero-shot services with a joint goal accuracy of 0.8652 and a slot tagging F1-Score of 0.9835.
- Open Access
Multi-Module System for Open Domain Chinese Question Answering over Knowledge Base
For the task of open domain Knowledge Based Question Answering in CCKS2019, we propose a method combining information retrieval and semantic parsing. This multi-module system extracts the topic entity and the most related relation predicate from a question and transforms it into a Sparql query statement. Our method obtained the F1 score of 70.45% on the test data.
Open AccessMulti-Module System for Open Domain Chinese Question Answering over Knowledge Base
For the task of open domain Knowledge Based Question Answering in CCKS2019, we propose a method combining information retrieval and semantic parsing. This multi-module system extracts the topic entity and the most related relation predicate from a question and transforms it into a Sparql query statement. Our method obtained the F1 score of 70.45% on the test data.
- 1 April 2013
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15253511,p. 256-261
Dynamic resource allocation for machine-type communications in LTE/LTE-A with contention-based access
In this paper, we propose a dynamic resource allocation method to enable efficient and low-latency machine type communications (MTC) in LTE/LTE-A with the contention based random access (CBA) scheme [9]. In the proposed method, we firstly estimate the probabilities of events caused by a CBA transmission and then calculate the latency with the measured resource unit. We increase the amount of CBA resources until the estimated latency satisfies the application QoS requirement. The simulation results demonstrate that with the proposed resource allocation method for CBA, the uplink channel access latency has been drastically reduced and that it always guarantees the latency requirements. Furthermore, the achievable latency is significantly reduced when compared to the regular scheduling and the standard random access scheme.
- 1 April 2013
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15253511,p. 256-261
Dynamic resource allocation for machine-type communications in LTE/LTE-A with contention-based access
In this paper, we propose a dynamic resource allocation method to enable efficient and low-latency machine type communications (MTC) in LTE/LTE-A with the contention based random access (CBA) scheme [9]. In the proposed method, we firstly estimate the probabilities of events caused by a CBA transmission and then calculate the latency with the measured resource unit. We increase the amount of CBA resources until the estimated latency satisfies the application QoS requirement. The simulation results demonstrate that with the proposed resource allocation method for CBA, the uplink channel access latency has been drastically reduced and that it always guarantees the latency requirements. Furthermore, the achievable latency is significantly reduced when compared to the regular scheduling and the standard random access scheme.
- 1 August 2008
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 870-875
Effects of synchronization errors on the performance of TDD/OFDMA crossed timeslot
As a key technology in future mobile communications system, OFDMA has efficient frequency utilization and strong robustness to frequency selective fading, however it's sensitive to synchronization errors. Although effects of synchronization errors on crossed timeslot's performance have been widely investigated in TDD/CDMA system, little work has done in TDD/OFDMA system. This paper analyzes the effects of synchronization errors on the performance of crossed timeslot in TDD/OFDMA system. It's proved that crossed timeslot is more sensitive to synchronization errors comparing with non-crossed timeslot in TDD/OFDMA system, and even very small frequency synchronization errors or sample clock frequency errors may bring serious degradation to system performance, to which attention should be paid in system design. The downlink user equipment (UE) in crossed timeslot receives strong interference when synchronization errors exist and the main part of the interference is caused by the uplink signal of other UEs in the crossed timeslot.
- 1 August 2008
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 870-875
Effects of synchronization errors on the performance of TDD/OFDMA crossed timeslot
As a key technology in future mobile communications system, OFDMA has efficient frequency utilization and strong robustness to frequency selective fading, however it's sensitive to synchronization errors. Although effects of synchronization errors on crossed timeslot's performance have been widely investigated in TDD/CDMA system, little work has done in TDD/OFDMA system. This paper analyzes the effects of synchronization errors on the performance of crossed timeslot in TDD/OFDMA system. It's proved that crossed timeslot is more sensitive to synchronization errors comparing with non-crossed timeslot in TDD/OFDMA system, and even very small frequency synchronization errors or sample clock frequency errors may bring serious degradation to system performance, to which attention should be paid in system design. The downlink user equipment (UE) in crossed timeslot receives strong interference when synchronization errors exist and the main part of the interference is caused by the uplink signal of other UEs in the crossed timeslot.