Si Pang

Mr. Si Pang


Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 July 2024 in Electronics
Verified
Journal Article
Open access
PDF file
HSHan Sun
Han Sun
CTChaohong Tan
Chaohong Tan
SPSi Pang
Si Pang
HWHancheng Wang
Hancheng Wang
BHBaohua Huang
Baohua Huang
To detect text from electronic seals that have significant background interference, blurring, text overlapping, and curving, an improved YOLOv8 model named RA-YOLOv8 was developed. The model is primarily based on YOLOv8, with optimized structures in its backbone and neck. The receptive-field attention and efficient multi-scale attention (RFEMA) module is introduced in the backbone. The model’s ability to extract and integrate local and global features is enhanced by combining the attention on the receptive-field spatial feature of the receptive-field attention and coordinate attention (RFCA) module and the cross-spatial learning of the efficient multi-scale attention (EMA) module. The Alterable Kernel Convolution (AKConv) module is incorporated in the neck, enhancing the model’s detection accuracy of curved text by dynamically adjusting the sampling position. Furthermore, to boost the model’s detection performance, the original loss function is replaced with the bounding box regression loss function of Minimum Point Distance Intersection over Union (MPDIoU). Experimental results demonstrate that RA-YOLOv8 surpasses YOLOv8 in terms of precision, recall, and F1 value, with improvements of 0.4%, 1.6%, and 1.03%, respectively, validating its effectiveness and utility in seal text detection.

13 May 2025 in Electronics
Verified
Journal Article
Open access
PDF file
SPSi Pang
Si Pang
JWJing Wen
Jing Wen
SLShaoling Liang
Shaoling Liang
BHBaohua Huang
Baohua Huang
Recent advancements in cloud computing, edge computing, and Internet of Things (IoT) have increased the complexity of network environments and provided fertile ground for malicious attacks. Existing DL-based malware detections, while making progress in detection accuracy and generalization ability, face serious challenges in user data privacy protection. To address this problem, this paper proposed a non-interactive malware detection system based on CKKS homomorphic encryption (FICConvNet). The system effectively achieves end-to-end data privacy protection, ensures that sensitive data uploaded by users are processed in an encrypted state, prevents data leakage, and protects the privacy of detection results. The key technology of FICConvNet is its innovative lightweight ciphertext inference architecture, which combines DS Conv and structured sparse projection to significantly reduce the complexity of homomorphic computation. Meanwhile, in this paper, an adaptive learnable activation function (ALPolyAct) is designed to replace the traditional fixed polynomial activation function to enhance the expressive power and inference accuracy of the model. In addition, the privacy protection of user data and the security of detection results are optimized by the zero-decryption inference process. Experimental results show that FICConvNet achieves a detection accuracy of 95.86%, which significantly outperforms the existing ciphertext inference model CryptoNets (15.5% improvement) and approaches the performance of the plaintext model ResNet-18. In addition, FICConvNet reduces ciphertext inference time by about 80% compared to Conv2d structures. The research in this paper provides an effective privacy-preserving solution in the field of malware detection and explores new research directions for the application of homomorphic encryption in malware detection.

9 February 2024 in Applied Sciences
Verified
Journal Article
Open access
PDF file
BHBaohua Huang
Baohua Huang
YLYunjie Lin
Yunjie Lin
SPSi Pang
Si Pang
LFLong Fu
Long Fu
Named entity recognition of government audit text is a key task of intelligent auditing. Aiming at the problems of scarcity of corpus in the field of governmental auditing, insufficient utilization of traditional character vector word-level information features, and insufficient capturing of auditing entity features, this study builds its own dataset in the field of auditing and proposes the model CW-CBGC for recognizing named entities in governmental auditing text based on ChineseBERT and character-word fusion. First, the ChineseBERT pre-training model is used to extract the character vector that integrates the features of glyph and pinyin, combining with word vectors dynamically constructed by the BERT pre-training model; then, the sequences of character-word fusion vectors are input into the bi-directional gated recurrent neural network (BiGRU) to learn the textual features. Finally, the global optimal sequence label is generated by Conditional Random Field (CRF), and the GHM classification loss function is used in the model training to solve the problem of error evaluation under the conditions of noisy entities and unbalanced number of entities. The F1 value of this study’s model on the audit dataset is 97.23%, which is 3.64% higher than the baseline model’s F1 value; the F1 value of the model on the public dataset Resume is 96.26%, which is 0.73–2.78% higher than the mainstream model. The experimental results show that the model proposed in this paper can effectively recognize the entities in government audit texts and has certain generalization ability.