HUANG TAI SHENG

Mr. HUANG TAI SHENG

Beijing Institute of Technology, Beijing, China
orcid

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

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

Representative Publications

View All Publications
1 December 2024 in Data Intelligence
Verified
Journal Article
Open access
PDF file
PJPang Jinhui
Pang Jinhui
XYXinyun Yanga
Xinyun Yanga
XQXiaoyao Qiub
Xiaoyao Qiub
ZWZixuan Wanga
Zixuan Wanga
HUANG TAI SHENGHUANG TAI SHENG
HUANG TAI SHENG
Multi-modal Named Entity Recognition(MNER), which is vision-language task, utilizes images as auxiliary to detect and classify named entities from input sentence. Recent studies find visual information is helpful for Named Entity Recognition, while the difference between those two modalities is not carefully considered. Therefore, these approaches utilizing different pre-trained models do not reduce the gap between textual and visual features, which give the same weight of different modalities usually predict wrong because of the noise of visual information. To reduce these bias, we propose a Masked Multi-modal Attention Fusion approach for MNER, named MMAF. Firstly, we utilize Image Caption to generate textual representation of image, which is combined with original sentence. Then, to get textual and visual features, we map the multi-modal inputs into a shared space and stack Multi-modal Attention Fusion layer that performs fully interaction between two modalities. We add Multi-modal Attention Mask to highlight the importance of certain words in sentences, enhancing the performance of entity detection. Finally, we achieve Multi-modal Attention based representation for each word and perform entity labeling via CRF decoder. Experiments show our method outperforms state-of-the-art models by 0.23% and 0.84% on Twitter 2015 and 2017 MNER datasets respectively, demonstrating its effectiveness.

24 May 2022 in Mathematics
Verified
Journal Article
Open access
PDF file
LLLang Liu
Lang Liu
LWLulu Wang
Lulu Wang
HUANG TAI SHENGHUANG TAI SHENG
HUANG TAI SHENG
PJPang Jinhui
Pang Jinhui
The dynamic optimization of the closed-loop supply chain (CLSC) is a hot research topic. Members’ competitive behavior and product goodwill play an important role in the decision making of CLSC members. In this paper, a closed-loop supply chain (CLSC) with competitive manufacturers and single retailers is studied, in which the manufacturer produces and recycles the products, and the retailer is responsible for the sales of the products. On this basis, a dynamic linear differential equation of product goodwill is constructed, the optimal dynamic path of each decision variable is found, and the influence of price competition among manufacturers on the decision making of members in a dynamic closed-loop supply chain is studied. The conclusion is verified by an example. The results show that goodwill directly affects the wholesale price, the retail price, the recovery price, and the profit of supply chain members. The wholesale price and the retail price of products are not only positively affected by their own goodwill, but also by the goodwill of competing products. The manufacturer competition intensity will affect the product price and the supply chain member profit. To a certain extent, the more intense the manufacturer’s competition is, the higher the wholesale price and the retail price, and the greater the profit of the supply chain members.

28 October 2024
Verified
Conference Paper
Open access
JPJin-hui PANG
Jin-hui PANG
CLChangqing Lin
Changqing Lin
XHXiaoshuai Hao
Xiaoshuai Hao
Rong YinRong Yin
Rong Yin
ZWZixuan Wang
Zixuan Wang
HUANG TAI SHENGHUANG TAI SHENG
HUANG TAI SHENG
Continual graph learning (CGL) is an important and challenging task that aims to extend static GNNs to dynamic task flow scenarios. As one of the mainstream CGL methods, the experience replay (ER) method receives widespread attention due to its superior performance. However, existing ER methods focus on identifying samples by feature significance or topological relevance, which limits their utilization of comprehensive graph data. In addition, the topology-based ER methods only consider local topological information and add neighboring nodes to the buffer, which ignores the global topological information and increases memory overhead. To bridge these gaps, we propose a novel method called Feature-Topology Fusion-based Experience Replay (FTF-ER) to effectively mitigate the catastrophic forgetting issue with enhanced efficiency. Specifically, from an overall perspective to maximize the utilization of the entire graph data, we propose a highly complementary approach including both feature and global topological information, which can significantly improve the effectiveness of the sampled nodes. Moreover, to further utilize global topological information, we propose Hodge Potential Score (HPS) as a novel module to calculate the topological importance of nodes. HPS derives a global node ranking via Hodge decomposition on graphs, providing more accurate global topological information compared to neighbor sampling. By excluding neighbor sampling, HPS significantly reduces buffer storage costs for acquiring topological information and simultaneously decreases training time. Compared with state-of-the-art methods, FTF-ER achieves a significant improvement of 3.6% in AA and 7.1% in AF on the OGB-Arxiv dataset, demonstrating its superior performance in the class-incremental learning setting.

13 March 2022 in Mathematics
Verified
Journal Article
Open access
PDF file
DHDonghong Huang
Donghong Huang
PJPang Jinhui
Pang Jinhui
LLLang Liu
Lang Liu
SWShuangsheng Wu
Shuangsheng Wu
HUANG TAI SHENGHUANG TAI SHENG
HUANG TAI SHENG
In the circumstance that unexpected events lead to the information asymmetry of sales costs, supplier risk aversion and stochastic price, this paper discusses the internal law of using an emergency quantity discount contract to coordinate the supply chain. First, the Conditional Value at Risk (CVaR) model of supplier risk aversion under the condition of information symmetry is constructed. In addition, the model is extended to the game model of the CVaR of supplier risk aversion under the condition of the asymmetric information of sales costs and solved. After that, the simulation test is performed. The results show that, firstly, under the condition of random price, the supplier risk aversion leads to the phenomenon of bifurcation and mutation in each decision variable of the supply chain system. Secondly, retailers can obtain excess profits by concealing private information, but this harms the interests of suppliers and the entire supply chain. Thirdly, suppliers with different risk attitudes should have different strategies concerning asymmetry in sales cost information. Fourthly, the more asymmetric the information for the sales costs, the more unstable the system. The conclusion is that the phenomenon of bifurcation mutation is the result of the coupling effect of price randomness and supplier risk aversion. The supply chain cannot coordinate in the bifurcation mutation region, but can coordinate outside of it. Hiding private information benefits those who own it, but harms the system as a whole.