Mingze Zheng
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Mingze Zheng has a total of 5 co-authors. Here are the top co-authors by number of shared publications:
- Changming Xu
School of Computer and Communication Engineering, Northeastern University, Qinhuangdao 066004, China
1 shared publication - Junwei Wang
School of Computer and Communication Engineering, Northeastern University, Qinhuangdao 066004, China
1 shared publication - Kuo Wu
School of Computer and Communication Engineering, Northeastern University, Qinhuangdao 066004, China
1 shared publication - Ranran Zhang
School of Computer and Communication Engineering, Northeastern University, Qinhuangdao 066004, China
1 shared publication - Xingcan Liu
School of Computer and Communication Engineering, Northeastern University, Qinhuangdao 066004, China
1 shared publication
Latest Publications
- Open Access
- 31 March 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (7) , 3395
Dotsformer: Capturing Chain-Loop Structures for Transformer in Dots-and-Boxes
In many board games, AlphaZero has demonstrated superhuman abilities. Dots-and-Boxes is a classic board game with simple rules but requiring skills to win. This paper proposes Dotsformer, which extracts chain-loop structures from the game board. These structures connect distant boxes, providing long-range relational information as input to the Transformer. We employ multiple convolutional kernels to generate Q, K, and V, and incorporate information about the box structure itself into the attention scores. We also incorporate auxiliary training tasks, including an initiative task and a classification task. These tasks determine whether to retain or relinquish the initiative in the current situation, and classify actions into forbidden, conceding, safe, and scoring moves. They provide additional supervisory signals and accelerate learning. The experimental results show that Dotsformer outperforms AlphaZero in both rollback speed and playing strength: it achieved a winning rate of 87.6% and an ELO rating lead of 340 points against the baseline. Additionally, ablation studies verify the effectiveness of each key module.
- 31 March 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (7) , 3395
Open AccessDotsformer: Capturing Chain-Loop Structures for Transformer in Dots-and-Boxes
In many board games, AlphaZero has demonstrated superhuman abilities. Dots-and-Boxes is a classic board game with simple rules but requiring skills to win. This paper proposes Dotsformer, which extracts chain-loop structures from the game board. These structures connect distant boxes, providing long-range relational information as input to the Transformer. We employ multiple convolutional kernels to generate Q, K, and V, and incorporate information about the box structure itself into the attention scores. We also incorporate auxiliary training tasks, including an initiative task and a classification task. These tasks determine whether to retain or relinquish the initiative in the current situation, and classify actions into forbidden, conceding, safe, and scoring moves. They provide additional supervisory signals and accelerate learning. The experimental results show that Dotsformer outperforms AlphaZero in both rollback speed and playing strength: it achieved a winning rate of 87.6% and an ELO rating lead of 340 points against the baseline. Additionally, ablation studies verify the effectiveness of each key module.