KDD 2022

last modify

Workshops

• 表示学习、不平衡学习：Deep learning practice and theory for high-dimensional, sparse, and imbalanced data

• 知识图谱：International Workshop on Knowledge Graphs: Open Knowledge Network

• 时间序列：Workshop on Mining and Learning from Time Series -- Deep Forecasting: Models, Interpretability, and Applications

论文

• Mask and Reason: Pre-Training Knowledge Graph Transformers for Complex Logical Queries

  • 知识图谱；KGTransformer；

  • 面向复杂逻辑查询的知识图谱Transformer预训练。

• AutoFAS: Automatic Feature and Architecture Selection for Pre-Ranking System

  • 搜索；粗排；双塔模型优化；神经网络框架搜索(Neural Architecture Search)

• Practical Counterfactual Policy Learning for Top-K Recommendations

  • 数据优化；

  • 问题：对于训练机器学习模型，一项关键任务是通过收集的反馈（例如，评分、点击）来构建训练数据。 然而，从理论和实际经验中可以发现，收集的反馈中选择偏差会导致训练得到的模型有偏，从而导致训练结果是不是最优策略。

• Applying Deep Learning Based Probabilistic Forecasting to Food Preparation Time for On-Demand Delivery Service

  • 配送时间预测；

  • 应用概率估计刻画订单出餐时间的不确定性（区别于点估计）；提出了S-QL损失函数（基于S-CRPS）

• A Framework for Multi-stage Bonus Allocation in meal delivery Platform

  • 应用；物流配送相关；

  • 多阶段送餐奖励框架；优化无人接单问题；

• DDEN:A Heterogeneous Learning-to-Rank Approach with Deep Debiasing Experts Network

  • 应用；异构排序；