本研究提出一套專為體育對決預測任務設計的深度學習架構，結合貝式技能更新機制、特徵交互建模與時序特徵處理，有效強化模型對選手能力動態變化與非遞移性效應的表徵能力。核心方法包括貝氏後驗更新以追蹤選手能力浮動與不確定性，特徵交互網路結合指數移動平均（EMA）特徵，以捕捉非遞移性效應並強化模型對當下賽局的判斷能力。
為進一步提高模型的穩健性與泛化能力（generalization ability），本研究採用預訓練凍結骨幹網路（frozen backbone）策略，以獲取穩定表徵後進行整合層微調，降低對特定模組的依賴。實驗結果顯示，所提方法在多項體育競技對決資料集上顯著優於傳統對決模型，展現了貝式推論與深度神經網路在體育對決預測上的整合潛力。

This study proposes a deep learning framework specifically designed for sports matchup prediction tasks. The framework integrates Bayesian skill updating, feature interaction modeling, and temporal feature processing to improve the model’s capacity to capture dynamic variations in athlete performance and intransitivity effects. Methods include Bayesian posterior updates to capture fluctuations and uncertainty in player states, and a feature interaction network augmented with exponential moving average (EMA) features to capture intransitivity effects while enhancing the model’s judgment in current matchups.
To further improve model robustness and generalization ability, we adopt a frozen backbone training strategy. This allows stable representation learning before fine-tuning the integration layers, thereby reducing dependency on specific components. Experimental results demonstrate that the proposed method significantly outperforms traditional matchup models across multiple sports datasets, highlighting the integration potential of Bayesian inference and deep neural networks in sports prediction tasks.

Blundell, C., Cornebise, J., Kavukcuoglu, K., and Wierstra, D. (2015). Weight uncertainty in neural networks. In Proceedings of the 32nd International Conference on Machine Learning, pages 1613–1622. PMLR.
Bradley, R. A. and Terry, M. E. (1952). Rank analysis of incomplete block designs: I. the method of paired comparisons. Biometrika, 39(3/4):324–345.
Chen, S. and Joachims, T. (2016a). Modeling intransitivity in matchup and comparison data. In Proceedings of the Ninth ACM International Conference on Web Search and Data Mining, pages 227–236.
Chen, S. and Joachims, T. (2016b). Predicting matchups and preferences in context. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 775–784.
Glorot, X. and Bengio, Y. (2010). Understanding the difficulty of training deep feedforward neural networks. In Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, pages 249–256. JMLR Workshop and Conference Proceedings.
Graves, A. (2011). Practical variational inference for neural networks. Advances in neural information processing systems, 24.
Gu, Y., Liu, Q., Zhang, K., Huang, Z., Wu, R., and Tao, J. (2021). Neuralac: Learning cooperation and competition effects for match outcome prediction. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 35, pages 4072–4080.
Herbrich, R., Minka, T., and Graepel, T. (2006). Trueskill™: a bayesian skill rating system. Advances in neural information processing systems, 19.
Li, M., Wu, J., Wang, X., Chen, C., Qin, J., Xiao, X., Wang, R., Zheng, M., and Pan, X. (2023). Aligndet: Aligning pre-training and fine-tuning in object detection. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 6866–6876.
Loshchilov, I. and Hutter, F. (2017). Decoupled weight decay regularization. arXiv
preprint arXiv:1711.05101.
Lowe, Z. (2017). Why the lebron-kyrie pick-and-roll is the deadliest weapon in the nba.
Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., Desmaison, A., Köpf, A., Yang, E. Z., DeVito, Z., Raison, M., Tejani, A., Chilamkurthy, S., Steiner, B., Fang, L., Bai, J., and Chintala, S. (2019). Pytorch: An imperative style, high-performance deep learning library. CoRR, abs/1912.01703.
Rusu, A. A., Rabinowitz, N. C., Desjardins, G., Soyer, H., Kirkpatrick, J., Kavukcuoglu, K., Pascanu, R., and Hadsell, R. (2016). Progressive neural networks. arXiv preprint arXiv:1606.04671.
Seabold, S. and Perktold, J. (2010). Statsmodels: Econometric and statistical modeling with python. In Proceedings of the 9th Python in Science Conference.