Most algorithms for solving POMDPs itera­ tively improve a value function that implic­ itly represents a policy and are said to search in value function space. This paper presents an approach to solving POMDPs that repre­ sents a policy explicitly as a finite-state con­ troller and iteratively improves the controller by search in policy space. Two related al­ gorithms illustrate this approach. ...

Modeling policies in reproducing kernel Hilbert space (RKHS) renders policy gradient reinforcement learning algorithms non-parametric. As a result, the policies become very flexible and have a rich representational potential without a predefined set of features. However, their performances might be either non-covariant under reparameterization of the chosen kernel, or very sensitive to step-siz...

model, then there should not exist a policy that leads from s′ to s in the original

Much recent work in reinforcement learning and stochastic optimal control has focused on algorithms that search directly through a space of policies rather than building approximate value functions. Policy search has numerous advantages: it does not rely on the Markov assumption, domain knowledge may be encoded in a policy, the policy may require less representational power than a value-functio...

This paper describes part of a solution to the interpretation of human-readable policy documents into semi-automatic conformance checking. Using a socio-cognitively motivated representation of shared knowledge, and applying appropriate inference mechanisms from a normative perspective, a mechanism to automatically detect potentially non-conforming blog entries is detailed. Candidate non-conform...

It is a well-documented fact that transnational corporations engaged in the production and distribution of health-harmful commodities have been able to steer policy approaches to address the associated burden of non-communicable diseases (NCDs). While the political influence that corporations wield stems in part from significant financial resources, it has also been ena...

When social resources are limited, improving the lot of the underprivileged comes at the expense of others. Thus, policies such as Affirmative Action (AA) – designed to increase the representation of minority people in higher education or employment – implicitly entail tradeoffs between groups. We propose that, while aversion to personor group-tradeoffs of this sort is widespread, the identifia...

This paper considers uncertain dynamic systems in which sensing is performed upon request. We introduce the concept of an active sensing policy which evaluates the current state information and determines whether or not to request sensing information. The sensing policy considered has the goal of bounding the state uncertainty in a given direction in the state space. The policy presented is sho...

We understand and respect the points raised by Rummel and Conley (2017) in response to our initial Forum Article (Fairén et al., 2017), which we acknowledge are informed and literate. Unfortunately, they are also unconvincing. Their comments clearly illustrate why we are not searching for life on Mars today and why we haven’t done so during the last decades. Hereafter, we respond point by point...