Issues of environment and environmental health involve multiple interests regarding e.g. political, societal, economical, and public concerns represented by different kinds of organizations and individuals. Not surprisingly, stakeholder and public participation has become a major issue in environmental and environmental health policy and assessment. The need for participation has been discussed...

BACKGROUND Development of health policy is a complex process that does not necessarily follow a particular format and a predictable trajectory. Therefore, agenda setting and selecting of alternatives are critical processes of policy development and can give insights into how and why policies are made. Understanding why some policy issues remain and are maintained whiles others drop off the agen...

As an important approach to solving complex sequential decision problems, reinforcement learning (RL) has been widely studied in the community of artificial intelligence and machine learning. However, the generalization ability of RL is still an open problem and it is difficult for existing RL algorithms to solve Markov decision problems (MDPs) with both continuous state and action spaces. In t...

Many continuous control tasks have bounded action spaces and clip out-of-bound actions before execution. Policy gradient methods often optimize policies as if actions were not clipped. We propose clipped action policy gradient (CAPG) as an alternative policy gradient estimator that exploits the knowledge of actions being clipped to reduce the variance in estimation. We prove that CAPG is unbias...

This article studies the robustness of policy iteration in context continuous-time infinite-horizon linear quadratic regulator (LQR) problem. It is shown that Kleinman's algorithm small-disturbance input-to-state stable, a property stronger than Sontag's local stability but weaker global stability. More precisely, whenever error each bounded and small, solutions are also enter small neighborhoo...