Birdie Blog

Thinking will not overcome fear but action will.

Neural Combinatorial Optimization with Heavy Decoder-Toward Large Scale Generalization

NIPS23 重解码轻编码 + 迭代重构局部解改进

Neural Combinatorial Optimization with Heavy Decoder: Toward Large Scale Generalization 重解码器的神经组合优化:面向大规模泛化 来自南方科技大学和香港城市大学 开源：https://github.com/CIAM-Group/NCO_code/tree/main/single_objective/L...

Posted by Birdie on January 23, 2024

Let the Flows Tell Solving Graph Combinatorial Problems with GFlowNets

NIPS23 GFlowNets

NIPS23 Let the Flows Tell Solving Graph Combinatorial Problems with GFlowNets 用GFlowNets求解图组合优化问题 来自Mila实验室和Google DeepMind 开源：https://github.com/zdhNarsil/GFlowNet-CombOpt （没研究过相关内容，没看懂） 摘要 ...

Posted by Birdie on January 23, 2024

双层规划问题

综述和一些应用

双层规划问题 \[\begin{aligned} \min_{(x,y)}\quad &f(x,y) \\ \mathrm{s.t.}\quad &g(x,y)\leqslant0\\ \quad &y\in S(x), \end{aligned}\] 其中 $S(x)$ 表示下层问题： \[\begin{aligned} \min_{y}\quad &a...

Posted by Birdie on January 22, 2024

Learning to Search Feasible and Infeasible Regions of Routing Problems with Flexible Neural k-Opt

NIPS23 k-opt learn to imporve + mask优化不可行探索 + 双流解码器

Learning to Search Feasible and Infeasible Regions of Routing Problems with Flexible Neural k-Opt 用柔性神经k-Opt学习搜索路由问题的可行和不可行区域 开源：https://github.com/yining043/NeuOpt 摘要 在本文中，我们提出了神经k-Opt (NeuO...

Posted by Birdie on January 21, 2024

DeepACO Neural-enhanced Ant Systems for Combinatorial Optimization

NIPS23 神经蚁群系统 构造+局部搜索 泛用的元启发式

DeepACO Neural-enhanced Ant Systems for Combinatorial Optimization DeepACO:用于组合优化的神经增强蚁群系统 代码：https://github.com/henry-yeh/DeepACO 摘要 蚁群优化算法是一种元启发式算法，已成功地应用于各种组合优化问题。传统上，针对特定问题定制蚁群算法需要知识驱动的启发...

Posted by Birdie on January 20, 2024

Ensemble-based Deep Reinforcement Learning for Vehicle Routing Problems under Distribution Shift

NIPS23 集成学习提高泛化性

Ensemble-based Deep Reinforcement Learning for Vehicle Routing Problems under Distribution Shift 配送移位下基于集成的车辆路径问题深度强化学习 摘要 虽然在独立同分布情况下表现良好，但大多数现有的VRP神经方法在分布变化的情况下难以泛化。为了解决这个问题，我们提出了一种基于集成的vrp深...

Posted by Birdie on January 19, 2024

Deep Reinforcement Learning for the Electric Vehicle Routing Problem With Time Windows

EVRPTW

Deep Reinforcement Learning for the Electric Vehicle Routing Problem With Time Windows 文章原文：https://ieeexplore.ieee.org/document/9520134 发表在2022 IEEE Transactions on Intelligent Transportation Sy...

Posted by Birdie on January 18, 2024

Combinatorial Optimization with Policy Adaptation using Latent Space Search

NIPS23 潜在空间采样学习+推理时候搜索

NIPS23 Combinatorial Optimization with Policy Adaptation using Latent Space Search 基于潜在空间搜索的策略自适应组合优化 来自InstaDeep 代码和数据集： https://github.com/instadeepai/compass （我没读代码，理解的不是很透彻，没有附录感觉讲的也不是很清楚，需...

Posted by Birdie on January 17, 2024

Winner Takes It All-Training Performant RL Populations for Combinatorial Optimization

NIPS23 多智能体（种群）学习

Winner Takes It All: Training Performant RL Populations for Combinatorial Optimization 赢家通吃:训练用于组合优化的高性能RL种群 来自InstaDeep 摘要 将强化学习(RL)应用于组合优化问题是有吸引力的，因为它不需要专家知识或预先解决的实例。然而，由于其固有的复杂性，期望智能体在一次推理中解...

Posted by Birdie on January 16, 2024

BQ-NCO Bisimulation Quotienting for Efficient Neural Combinatorial Optimization

NIPS23 新的马尔可夫决策过程表示并改进学习框架

BQ-NCO Bisimulation Quotienting for Efficient Neural Combinatorial Optimization BQ-NCO:高效神经组合优化的双模拟分法（Bisimulation Quotienting不知道中文怎么翻译） 来自Naver Labs Europe 代码：naver/bq-nco (github.com) 摘要 尽管基...

Posted by Birdie on January 16, 2024

FEATURED TAGS
神经网络 机器学习 机器人 webot 通信原理 并行计算 信息安全 自然语言处理 数据库 数字图像处理 软件工程 编译原理 强化学习 数字电路 proteus 计算机组成原理 VIVADO JAVA 论文阅读 组合优化 NIPS AAAI ICML IJCAI 隐私保护 ICLR
ABOUT ME

Birdie-Go

✉️ fangzhh6@mail2.sysu.edu.cn