Search

A dynamic pricing problem is difficult due to the highly dynamic environment and unknown demand distributions. In this article, we propose a deep reinforcement learning (DRL) framework, which is a pipeline that automatically defines the DRL components for solving a dynamic pricing problem. The automated DRL pipeline is necessary because the DRL framework can be designed in numerous ways, and manually finding optimal configurations is tedious. The levels of automation make nonexperts capable of using DRL for dynamic pricing. Our DRL pipeline contains three steps of DRL design, including Markov decision process modeling, algorithm selection, and hyperparameter optimization. It starts with transforming available information to state representation and defining reward function using a reward shaping approach. Then, the hyperparameters are tuned using a novel hyperparameter optimization method that integrates Bayesian optimization and the selection operator of the genetic algorithm. We employ our DRL pipeline on reserve price optimization problems in online advertising as a case study. We show that using the DRL configuration obtained by our DRL pipeline, a pricing policy is obtained whose revenue is significantly higher than the benchmark methods. The evaluation is performed by developing a simulation for the real-time bidding environment that makes exploration possible for the reinforcement learning agent.

Presents information on the 2015 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE 2015).

We consider conditional density approximation by fuzzy systems. Fuzzy systems are typically used to approximate deterministic functions in which the stochastic uncertainty is ignored. We propose probabilistic fuzzy systems (PFSs), in which the probabilistic nature of uncertainty is taken into account. These systems take also fuzzy uncertainty into account by their fuzzy partitioning of input and output spaces. We discuss an additive reasoning scheme for PFSs that leads to the estimation of conditional probability densities and prove how such fuzzy systems compute the expected value of this conditional density function. We show that some of the most commonly used fuzzy systems can compute the same expected output value, and we derive how their parameters should be selected in order to achieve this goal. The additional information and process understanding provided by the different interpretations of the PFS models are illustrated using a real-world example.

In this paper, we visualize the structure and the evolution of the computational intelligence (CI) field. Based on our visualizations, we analyze the way in which the CI field is divided into several subfields. The visualizations provide insight into the characteristics of each subfield and into the relations between the subfields. By comparing two visualizations, one based on data from 2002 and one based on data from 2006, we examine how the CI field has evolved over the last years. A quantitative analysis of the data further identifies a number of emerging areas within the CI field. The data that we use consist of the abstracts of the papers presented at the IEEE World Congress on Computational Intelligence (WCCI) in 2002 and 2006. Using a fully automatic procedure, so-called concept maps are constructed from the data. These maps visualize the associations between the main concepts in the CI field. Our analysis of the structure and the evolution of the CI field are largely based on the constructed concept maps