Random propagation in complex systems : nonlinear matrix recursions and epidemic spread

This dissertation studies long-term behavior of random Riccati recursions and mathematical epidemic model. Riccati recursions are derived from Kalman filtering. The error covariance matrix of Kalman filtering satisfies Riccati recursions. Convergence condition of time-invariant Riccati recursions are well-studied by researchers. We focus on time-varying case, and assume that regressor matrix is random and identical and independently distributed according to given distribution whose probability distribution function is continuous, supported on whole space, and decaying faster than any polynomial. We study the geometric convergence of the probability distribution. We also study the global dynamics of the epidemic spread over complex networks for various models. For instance, in the discrete-time Markov chain model, each node is either healthy or infected at any given time. In this setting, the number of the state increases exponentially as the size of the network increases. The Markov chain has a unique stationary distribution where all the nodes are healthy with probability 1. Since the probability distribution of Markov chain defined on finite state converges to the stationary distribution, this Markov chain model concludes that epidemic disease dies out after long enough time. To analyze the Markov chain model, we study nonlinear epidemic model whose state at any given time is the vector obtained from the marginal probability of infection of each node in the network at that time. Convergence to the origin in the epidemic map implies the extinction of epidemics. The nonlinear model is upper-bounded by linearizing the model at the origin. As a result, the origin is the globally stable unique fixed point of the nonlinear model if the linear upper bound is stable. The nonlinear model has a second fixed point when the linear upper bound is unstable. We work on stability analysis of the second fixed point for both discrete-time and continuous-time models. Returning back to the Markov chain model, we claim that the stability of linear upper bound for nonlinear model is strongly related with the extinction time of the Markov chain. We show that stable linear upper bound is sufficient condition of fast extinction and the probability of survival is bounded by nonlinear epidemic map.

Natural actor and belief critic: Reinforcement algorithm for learning parameters of dialogue systems modelled as POMDPs

This article presents a novel algorithm for learning parameters in statistical dialogue systems which are modeled as Partially Observable Markov Decision Processes (POMDPs). The three main components of a POMDP dialogue manager are a dialogue model representing dialogue state information; a policy that selects the system's responses based on the inferred state; and a reward function that specifies the desired behavior of the system. Ideally both the model parameters and the policy would be designed to maximize the cumulative reward. However, while there are many techniques available for learning the optimal policy, no good ways of learning the optimal model parameters that scale to real-world dialogue systems have been found yet. The presented algorithm, called the Natural Actor and Belief Critic (NABC), is a policy gradient method that offers a solution to this problem. Based on observed rewards, the algorithm estimates the natural gradient of the expected cumulative reward. The resulting gradient is then used to adapt both the prior distribution of the dialogue model parameters and the policy parameters. In addition, the article presents a variant of the NABC algorithm, called the Natural Belief Critic (NBC), which assumes that the policy is fixed and only the model parameters need to be estimated. The algorithms are evaluated on a spoken dialogue system in the tourist information domain. The experiments show that model parameters estimated to maximize the expected cumulative reward result in significantly improved performance compared to the baseline hand-crafted model parameters. The algorithms are also compared to optimization techniques using plain gradients and state-of-the-art random search algorithms. In all cases, the algorithms based on the natural gradient work significantly better. © 2011 ACM.

Reinforcement learning for parameter estimation in statistical spoken dialogue systems

Reinforcement techniques have been successfully used to maximise the expected cumulative reward of statistical dialogue systems. Typically, reinforcement learning is used to estimate the parameters of a dialogue policy which selects the system's responses based on the inferred dialogue state. However, the inference of the dialogue state itself depends on a dialogue model which describes the expected behaviour of a user when interacting with the system. Ideally the parameters of this dialogue model should be also optimised to maximise the expected cumulative reward. This article presents two novel reinforcement algorithms for learning the parameters of a dialogue model. First, the Natural Belief Critic algorithm is designed to optimise the model parameters while the policy is kept fixed. This algorithm is suitable, for example, in systems using a handcrafted policy, perhaps prescribed by other design considerations. Second, the Natural Actor and Belief Critic algorithm jointly optimises both the model and the policy parameters. The algorithms are evaluated on a statistical dialogue system modelled as a Partially Observable Markov Decision Process in a tourist information domain. The evaluation is performed with a user simulator and with real users. The experiments indicate that model parameters estimated to maximise the expected reward function provide improved performance compared to the baseline handcrafted parameters. © 2011 Elsevier Ltd. All rights reserved.

Policy optimisation of POMDP-based dialogue systems without state space compression

The partially observable Markov decision process (POMDP) has been proposed as a dialogue model that enables automatic improvement of the dialogue policy and robustness to speech understanding errors. It requires, however, a large number of dialogues to train the dialogue policy. Gaussian processes (GP) have recently been applied to POMDP dialogue management optimisation showing an ability to substantially increase the speed of learning. Here, we investigate this further using the Bayesian Update of Dialogue State dialogue manager. We show that it is possible to apply Gaussian processes directly to the belief state, removing the need for a parametric policy representation. In addition, the resulting policy learns significantly faster while maintaining operational performance. © 2012 IEEE.

POMDP-based statistical spoken dialog systems: A review

Statistical dialog systems (SDSs) are motivated by the need for a data-driven framework that reduces the cost of laboriously handcrafting complex dialog managers and that provides robustness against the errors created by speech recognizers operating in noisy environments. By including an explicit Bayesian model of uncertainty and by optimizing the policy via a reward-driven process, partially observable Markov decision processes (POMDPs) provide such a framework. However, exact model representation and optimization is computationally intractable. Hence, the practical application of POMDP-based systems requires efficient algorithms and carefully constructed approximations. This review article provides an overview of the current state of the art in the development of POMDP-based spoken dialog systems. © 1963-2012 IEEE.

Local equilibrium of the Gibbs interface in two-phase systems

We analyze the local equilibrium assumption for interfaces from the perspective of gauge transformations, which are the small displacements of Gibbs' dividing surface. The gauge invariance of thermodynamic properties turns out to be equivalent to conditions for jumps of bulk densities across the interface. This insight strengthens the foundations of the local equilibrium assumption for interfaces and can be used to characterize nonequilibrium interfaces in a compact and consistent way, with a clear focus on gauge-invariant properties. Using the principle of gauge invariance, we show that the validity of Clapeyron equations can be extended to nonequilibrium interfaces, and an additional jump condition for the momentum density is recognized to be of the Clapeyron type. © 2012 Europhysics Letters Association.

On-line policy optimisation of Bayesian spoken dialogue systems via human interaction

A partially observable Markov decision process has been proposed as a dialogue model that enables robustness to speech recognition errors and automatic policy optimisation using reinforcement learning (RL). However, conventional RL algorithms require a very large number of dialogues, necessitating a user simulator. Recently, Gaussian processes have been shown to substantially speed up the optimisation, making it possible to learn directly from interaction with human users. However, early studies have been limited to very low dimensional spaces and the learning has exhibited convergence problems. Here we investigate learning from human interaction using the Bayesian Update of Dialogue State system. This dynamic Bayesian network based system has an optimisation space covering more than one hundred features, allowing a wide range of behaviours to be learned. Using an improved policy model and a more robust reward function, we show that stable learning can be achieved that significantly outperforms a simulator trained policy. © 2013 IEEE.