In light of the escalating popularity of online shopping and the urgent need to reduce unnecessary driving to mitigate environmental impacts, it has become increasingly important to provide cost-effective solutions for attended home deliveries. Extensive research efforts have been dedicated to addressing challenges related to integrating demand management and vehicle routing with time windows. In this paper, we present two key contributions. Firstly, we propose an enhanced method for estimating opportunity cost, by leveraging a dynamic-routing and distribution approach that incorporates forecast orders. This approach allows for more accurate revenue-loss assessments, ultimately leading to improved decision-making on the delivery charges. Secondly, we introduce a dynamic slot-combination strategy, aiming to fully exploit the flexibility that customers possess in receiving their delivery, which enhances overall route efficiency and customer satisfaction. Importantly, our proposed augmented time-windows approach can be easily implemented within existing systems, employing standard time windows, without necessitating any strategic changes or complex computational modifications to the routing system. To assess the performance of our proposed approach, we conducted exhaustive experiments on real data. The results demonstrate that our generated solution outperforms both recent and current state-of-the-art approaches in terms of profitability and delivery efficiency. This signifies the effectiveness and practicality of our proposed methodology in addressing the challenges associated with attended home deliveries.
M. Abdollahi, Xinan Yang, Michael Fairbank· Annals of Operations Researc...· 0 citations
The dominance of Neural Networks (NNs) in RL is partially due to their incremental learning capability, which naturally suits the online, non-stationary nature of self-play training. However, gradient-boosted trees like LightGBM are widely recognised as the state of the art for tabular data in supervised learning, often outperforming NNs in accuracy and efficiency. Game states are inherently tabular---discrete actions, categorical card identities, structured board positions---which makes them an ideal candidate for tree-based methods. We introduce LUGL (Local Updates, Global Learning), a framework that decouples data collection from model fitting, enabling non-incremental learners such as GBTs to operate in RL settings where they would otherwise fail due to distributional shift. LUGL alternates between a local updates phase, where the agent plays self-play games and accumulates tabular updates (Q-values, V-values, policies, or regret values) in a finite table, and a global learning phase, where the table is used to train a function approximator that generalises to unseen states before the table is reset. We test our approach in four standard perfect-information games (Tic-tac-toe, Connect-4, Othello, and Hex) and five imperfect-information games (Kuhn's poker, Leduc Hold'em, Liar's Dice, Goofspiel, and Flop5 Hold'em), and show that our results are competitive with or superior to DQN and DeepCFR. Our experiments demonstrate that the community's strong bias towards NNs in game-playing may be unwarranted, since LightGBM-based agents achieve competitive or superior performance across all tested benchmarks.
David Milec, Spyridon Samothrakis, Michael Fairbank et al.· IEEE Transactions on Games· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
