. Initially introduced in the late 1960s and early 1970s, dynamic programming algorithms have become increasingly popular in automatic speech recognition. There are two reasons why this has occurred: First, the dynamic programming strategy can be combined with avery e#cient and practical pruning strategy so that very large search spaces can be handled. Second, the dynamic programming strategy has turned out to be extremely #exible in adapting to new requirements. Examples of such requirements are the lexical tree organization of the pronunciation lexicon and the generation of a word graph instead of the single best sentence. In this paper, we attempt to systematically review the use of dynamic programming search strategies for small#vocabulary and large#vocabulary continuous speech recognition. The following methods are described in detail: search using a linear lexicon, search using a lexical tree, language-model look-ahead and word graph generation. 1 Introduction Search strategie...

Abstract–This paper proposes Adaptive Multi-Path routing (AMP) as a simple algorithm for dynamic traffic engineering within autonomous systems. In contrast to related multipath routing proposals, AMP does not employ a global perspective of the network in each node. It restricts available information to a local scope, which opens the potential of reducing signaling overhead and memory consumption in routers. Having implemented AMP in ns-2, the algorithm is compared to standard routing strategies for a realistic simulation scenario. The results demonstrate the stability of AMP as well as the significant performance gains achieved. I. INTRODUCTION AND RELATED WORK Efficient routing algorithms have always been among the core building blocks of any packet switching network. Whereas existing routing protocols are usually designed for

We are interested in describing timed systems that exhibit probabilistic behaviour. To this purpose, we consider a model of Probabilistic Timed Automata and introduce a concept of weak bisimulation for these automata, together with an algorithm to decide it. The weak bisimulation relation is shown to be preserved when either time, or probability are abstracted away. As an application, we use weak bisimulation for Probabilistic Timed Automata to model and analyze a timing attack on the dining cryptographers protocol.

Planning problems that involve learning a policy from a single training set of finite horizon trajectories arise in both social science and medical fields. We consider Q-learning with function approximation for this setting and derive an upper bound on the generalization error. This upper bound is in terms of quantities minimized by a Q-learning algorithm, the complexity of the approximation space and an approximation term due to the mismatch between Q-learning and the goal of learning a policy that maximizes the value function.

Valuation-Based Systems (VBS) for belief-functions theory is applied to Pavement Management Systems (PMS) decision-making. The VBS provides a general framework for representing knowledge and drawing inferences under uncertainty. A VBS network is constructed and potentials are introduced in the form of belief-function (or basic probability assignment) in PMS decision making environment. Valuation network is another method of representing and solving Bayesian decision problems. It is based on the framework of VBS. Valuation network depict decision variables, random variables, utility functions, and information constraints. The solution method for valuation network is called fusion algorithm, and the Dempster's rule of combination can be successfilly applied in this framework. It will be shown that this approach can capture the quantitative, qualitative and incomplete information in PMS decision making. 1.

Abstract—In multilevel systems it is important to avoid unwanted indirect information flow from higher levels to lower levels, namely the so called covert channels. Initial studies of information flow analysis were performed by abstracting away from time and probability. It is already known that systems that are proved to be secure in a possibilistic framework may turn out to be insecure when time or probability are considered. Recently, work has been done in order to consider also aspects either of time or of probability, but not both. In this paper we propose a general framework, based on Probabilistic Timed Automata, where both probabilistic and timing covert channels can be studied. We define a Non-Interference security property and a Non Deducibility on Composition security property, which allow expressing information flow in a timed and probabilistic setting. We then compare these properties with analogous ones defined in contexts where either time or probability or neither of them are taken into account. This permits a classification of the properties depending on their discerning power. As an application, we study a system with covert channels that we are able to discover by applying our techniques.