Abstract
Many episode-based (or direct) policy search algorithms, maintain a multivariate Gaussian distribution as search distribution over the parameter space of some objective function. One class of algorithms, such as episodic REPS, PoWER or PI2 uses, a weighted maximum likelihood estimate (WMLE) to update the mean and covariance matrix of this distribution in each iteration. However, due to high dimensionality of covariance matrices and limited number of samples, the WMLE is an unreliable estimator. The use of WMLE leads to overfitted covariance estimates, and, hence the variance/entropy of the search distribution decreases too quickly, which may cause premature convergence. In order to alleviate this problem, the estimated covariance matrix can be regularized in different ways, for example by using a convex combination of the diagonal covariance estimate and the sample covariance estimate. In this paper, we propose a new covariance matrix regularization technique for policy search methods that uses the convex combination of the sample covariance matrix and the old covariance matrix used in last iteration. The combination weighting is determined by specifying the desired entropy of the new search distribution. With this mechanism, the entropy of the search distribution can be gradually decreased without damage from the maximum likelihood estimate.

Abstract
The mainstream of embedded software development as of today is dominated by C programming. To aid the development, hardware abstractions, libraries, kernels and lightweight operating systems are commonplace. Such kernels and operating systems typically impose a thread based abstraction to concurrency. However, in general thread based programming is hard, plagued by race conditions and dead-locks. For this paper we take an alternative outset in terms of a language abstraction, RTFM-core, where the system is modelled directly in terms of tasks and resources. In compliance to the Stack Resource Policy (SRP) model, the language enforces (well-formed) LIFO nesting of claimed resources, thus SRP based analysis and scheduling can be readily applied. For the execution onto bare-metal single core architectures, the rtfm-core compiler performs SRP analysis on the model and render an executable that is deadlock free and (through RTFM-kernel primitives) exploits the underlying interrupt hardware for efficient scheduling. The RTFM-core language embeds C-code and links to C-object files and libraries, and is thus applicable to the mainstream of embedded development. However, while the language enforces well-formed resource management, control flow in the embedded C-code may violate the LIFO nesting requirement. In this paper we address this issue by lifting a subset of C into the RTFM-core language allowing arbitrary control flow at the model level. In this way well-formed LIFO nesting can be enforced, and models ensured to be correct by construction. We demonstrate the feasibility by means of a prototype implementation in the rtfm-core compiler. Additionally, we develop a set of running examples and show in detail how control flow is handled at compile time and during run-time execution. © 2015 IEEE.

Abstract
Project-based organizations have characteristics that raise additional barriers to information management, knowledge sharing, and to organizational learning. The main causes of this are inadequate information architectures and governance, poor collaborative culture, and lack of organization-wide information management strategies. This chapter presents a comprehensive basis to understand the information and knowledge-sharing practices in PBO, as well as the methods and tools that information professionals and project managers should have in mind when performing their tasks. For that, literatures are reviewed focusing on the explanation of the processes of knowledge creation and sharing leading to organizational learning. The main conclusion is that a knowledge-sharing strategy in a PBO should include a set of mechanisms that address a customized mix of the codification and personalization dimensions and that strategies for collaborative information management should be used as enablers for embedding knowledge sharing within the organizational practices and culture.

Abstract
Stochastic search algorithms are general black-box optimizers. Due to their ease of use and their generality, they have recently also gained a lot of attention in operations research, machine learning and policy search. Yet, these algorithms require a lot of evaluations of the objective, scale poorly with the problem dimension, are affected by highly noisy objective functions and may converge prematurely. To alleviate these problems, we introduce a new surrogate-based stochastic search approach. We learn simple, quadratic surrogate models of the objective function. As the quality of such a quadratic approximation is limited, we do not greedily exploit the learned models. The algorithm can be misled by an inaccurate optimum introduced by the surrogate. Instead, we use information theoretic constraints to bound the 'distance' between the new and old data distribution while maximizing the objective function. Additionally the new method is able to sustain the exploration of the search distribution to avoid premature convergence. We compare our method with state of art black-box optimization methods on standard uni-modal and multi-modal optimization functions, on simulated planar robot tasks and a complex robot ball throwing task. The proposed method considerably outperforms the existing approaches.

Abstract
The IEC 61499 standard provides means to specify distributed control systems in terms of function blocks. The execution model is event driven (asynchronous), where triggering events may be associated with data (and seen as a message). In this paper we propose a low complexity implementation technique allowing to assess end-to-end response time of event chains spanning over a set of networked devices. In this paper we develop a method to provide safe end-to-end response time taking both intra- and inter-device delivery delays into account. As a use case we study the implementation onto (single-core) ARM-cortex based devices communicating over a switched Ethernet network. For the analysis we define a generic switch model and an experimental setup allowing us to study the impact of network topology as well as 802.1Q quality of service in a mixed critical setting. Our results indicate that safe sub millisecond end-to-end response times can be obtained using the proposed approach. © 2015 IEEE.

Abstract
The high penetration of wind and photovoltaic power in electricity markets will represent a major challenge in the forthcoming years. The main problem of both technologies is the high uncertainty in their production and their dependence on environmental conditions. The coordination between wind and photovoltaic power aims to lower imbalances, reducing their associated penalties. This paper describes two strategies: i) separate wind and photovoltaic strategy and ii) single wind-photovoltaic strategy. The strategies proposed are solved through stochastic mixed integer linear programming. The expected profits are maximized and they are obtained by selling the energy in the day-ahead market. The imbalances are penalized in the balancing market as well. The model is tested for a week, 168 hours, and the data used come from the Spanish electricity market. The results of the case study are discussed, comparing both strategies. Following the discussion, the most important conclusions are presented.