<abstract><p>Graph neural networks (GNNs) is applied successfully in many graph tasks, but there still exists a limitation that many of GNNs model do not consider uncertainty quantification of its output predictions. For uncertainty quantification, there are mainly two types of methods which are frequentist and Bayesian. But both methods need to sampling to gradually approximate the real distribution, in contrast, evidential deep learning formulates learning as an evidence acquisition process, which could get uncertainty quantification by placing evidential priors over the original Gaussian likelihood function and training the NN to infer the hyperparameters of the evidential distribution without sampling. So evidential deep learning (EDL) has its own advantage in measuring uncertainty. We apply it with diffusion convolutional recurrent neural network (DCRNN), and do the experiment in spatiotemporal forecasting task in a real-world traffic dataset. And we choose mean interval scores (MIS), a good metric for uncertainty quantification. We summarized the advantages of each method.</p></abstract>

Product allocation in retail is the process of placing products throughout a store to connect consumers with relevant products. Discovering a good allocation strategy is challenging due to the scarcity of data and the high cost of experimentation in the physical world. Some work explores Reinforcement learning (RL) as a solution, but these approaches are often limited because of the sim2real problem. Learning policies from logged trajectories of a system is a key step forward for RL in physical systems. Recent work has shown that model-based offline RL can improve the effectiveness of offline policy estimation through uncertainty-penalized exploration. However, existing work assumes a continuous state space and access to a covariance matrix of the environment dynamics, which is not possible in the discrete case. To solve this problem, we propose a Bayesian model-based technique that naturally produces probabilistic estimates of the environment dynamics via the posterior predictive distribution, which we use for uncertainty-penalized exploration. We call our approach Posterior Penalized Offline Policy Optimization (PPOPO). We show that our world model better fits historical data due to informative priors, and that PPOPO outperforms other offline techniques in simulation and against real-world data.

No abstract

No abstract