Ziyue Zhang

Ziyue Zhang

Auto-generated profile, based on publicly available information

Ziyue Zhang

University of Electronic Science and Technology of China,School of Information and Software Engineering,Chengdu,China,610054

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

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

Representative Publications

View All Publications
25 November 2021 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
ZLZhipeng Luo
Zhipeng Luo
ZZZiyue Zhang
Ziyue Zhang
WLWen Li
Wen Li
YCYiping Chen
Yiping Chen
Cheng WangCheng Wang
Cheng Wang
Jonathan Li Jonathan Li
Jonathan Li
Individual tree detection is critical for forest investigation and monitoring. Several existing methods have difficulties to detect trees in complex forest environments due to insufficiently mining descriptive features. This study proposes a deep learning (DL) framework based on a designed multichannel information complementarity representation for detecting trees in complex forest using UAV laser scanning point clouds. The proposed method consists of two main stages: ground filtering and tree detection. In the first stage, a modified graph convolution network with a local topological information layer is designed to separate the ground points. Unlike most existing parametric methods, our ground filtering method avoids the optimal parameters selection to adapt to different kinds of environments. For tree detection, a top-down slice (TDS) module is first designed to mine the vertical structure information in a top-down way. Then, a special multichannel representation (MCR) is developed to preserve different distribution patterns of points from complementary perspectives. Finally, a multibranch network (MBNet) is proposed for individual tree detection by fusing multichannel features, which can provide discriminative information for MBNet to detect trees more accurately. MBNet was evaluated on seven forest areas [UAV light detection and ranging (LiDAR) data with the mean size of 14 $000~\text {m}^{2}$ and point density of 250 points/ $\text {m}^{2}$ ]. Experimental results showed that the proposed framework achieves excellent performance. Our method obtains promising performance with a mean recall of 89.23% and a mean F1-score of 87.04%.

8 December 2025 in Pattern Recognition
Verified
Journal Article
Closed access
ZZZiyue Zhang
Ziyue Zhang
MLMingbao Lin
Mingbao Lin
QSQuanjian Song
Quanjian Song
YZYuxin Zhang
Yuxin Zhang
RJRongrong Ji
Rongrong Ji

15 December 2023
Verified
Conference Paper
Closed access
ZZZiyue Zhang
Ziyue Zhang
GXGaojie Xiong
Gaojie Xiong
YYYongbin Yu
Yongbin Yu
XWXiangxiang Wang
Xiangxiang Wang
XFXiao Feng
Xiao Feng
NTNyima Tashi
Nyima Tashi
This paper highlights the low-resource Tibetan few-shot learning model and establishes accuracy benchmarks utilizing Matching Networks (MN). Addressing the issues of low quality and limited usability in the existing Tibetan-Chinese parallel corpus, this paper constructs a small-scale and high-quality Tibetan-Chinese parallel corpus containing 110,000 sentence pairs using data filtering, deduplication, deletion of blank lines, and special symbol processing. Based on full context embeddings, Softmax-based attention mechanisms, and similarity metrics, the Tibetan few-shot learning model based on MN is proposed to help the model learn logical Tibetan knowledge rapidly from few-shot data. Different distance metrics on Cosine, Euclidean, Poincare, and Minkowski distances are used to test the accuracy of the Tibetan k-shot learning model, which provides a unified and reliable benchmark for the Tibetan few-shot learning tasks. This research provides substantial assistance in solving the issues of machine learning performing poorly in low-resource settings. The experiment results demonstrate that the accuracy of all metrics is improved with the number of shots increased, and Euclidean distance performs well in 5-way 1-shot learning, cosine performs well in 5-way 2-shot learning, and Minkowski distance $(\mathrm{p}=3)$ shows excellent accuracy in 5-way 3-shot learning. Notably, we used our own Tibetan data set and succeeded in enhancing accuracy by a noteworthy 1.6% in a 5-way 3-shot setup.

15 June 2022
Verified
Conference Paper
Closed access
PDF file
YWYuan Wang
Yuan Wang
ZZZiyue Zhang
Ziyue Zhang
ZLZihao Li
Zihao Li
Maximizing returns is a constant pursuit of the financial market. In this paper, we show the relevant models we designed to realize the optimal allocation and combination of gold, bitcoin and cash. Advanced forecasting technology can promote the maturity of transactions. Our model takes the lead in introducing the rolling learning prediction method of feedforward multilayer neural network (MLP), which provides a method for accurately predicting market fluctuations from another way. We improved the original algorithm and changed the unified learning method to the rolling learning method based on the existing data before each day. Our practice has proved that our technology has better prediction effect for assets with less market volatility. For example, for gold, a safe haven asset, our prediction accuracy is 99.5%; For products with large market fluctuations, the prediction accuracy is relatively low. For example, for speculative assets with large market fluctuations such as bitcoin, our prediction accuracy is also as high as 97%.