Parameter learning for POMDP spoken dialogue models

The partially observable Markov decision process (POMDP) provides a popular framework for modelling spoken dialogue. This paper describes how the expectation propagation algorithm (EP) can be used to learn the parameters of the POMDP user model. Various special probability factors applicable to this task are presented, which allow the parameters be to learned when the structure of the dialogue is complex. No annotations, neither the true dialogue state nor the true semantics of user utterances, are required. Parameters optimised using the proposed techniques are shown to improve the performance of both offline transcription experiments as well as simulated dialogue management performance. ©2010 IEEE.

An overview of Sequential Monte Carlo methods for parameter estimation in general state-space models

Nonlinear non-Gaussian state-space models arise in numerous applications in control and signal processing. Sequential Monte Carlo (SMC) methods, also known as Particle Filters, provide very good numerical approximations to the associated optimal state estimation problems. However, in many scenarios, the state-space model of interest also depends on unknown static parameters that need to be estimated from the data. In this context, standard SMC methods fail and it is necessary to rely on more sophisticated algorithms. The aim of this paper is to present a comprehensive overview of SMC methods that have been proposed to perform static parameter estimation in general state-space models. We discuss the advantages and limitations of these methods. © 2009 IFAC.

Parameter measurement for heavy-vehicle fuel consumption modelling

A mathematical model is developed to predict the energy consumption of a heavy vehicle. It includes the important factors of heavy-vehicle energy consumption, namely engine and drivetrain performances, losses due to accessories, aerodynamic drag, rolling resistance, road gradients, and driver behaviour. Novel low-cost testing methods were developed to determine engine and drivetrain characteristics. A simple drive cycle was used to validate the model. The model is able to predict the fuel use for a 371 tractor-semitrailer vehicle over a 4 km drive cycle within 1 per cent. This paper demonstrates that accurate and reliable vehicle benchmarking and model parameter measurement can be achieved without expensive equipment overheads, e.g. engine and chassis dynamometers.

A framework for analysing business performance, firm innovation and related contextual factors: Perceptions of managers and policy makers in two European regions

The aim of this paper is to propose a novel reference framework that can be used to study how different kinds of innovation can result in better business performance and how external factors can influence both the firm's capacity to innovate and innovation itself. The value of the framework is demonstrated as it is applied in an exploratory study of the perceptions of public policy makers and managers from two European regions - the Veneto Region in Italy and the East of England in the UK. Amongst other things, the data gathered suggest that managers are generally less convinced than public policy makers, that the innovativeness of a firm is affected by factors over which policy makers have some control. This finding poses the question "what, if any, role can public policy makers play in enhancing a company's competitiveness by enabling it to become more innovative?".

Seismic model parameter estimation using least-squares migration

We describe a method for verifying seismic modelling parameters. It is equivalent to performing several iterations of unconstrained least-squares migration (LSM). The approach allows the comparison of modelling/imaging parameter configurations with greater confidence than simply viewing the migrated images. The method is best suited to determining discrete parameters but can be used for continuous parameters albeit with greater computational expense.

Parameter extraction and calorimetric validation for a silicon carbide jfet pspice model

This paper focuses on the PSpice model of SiC-JFET element inside a SiCED cascode device. The device model parameters are extracted from the I-V and C-V characterization curves. In order to validate the model, an inductive test rig circuit is designed and tested. The switching loss is estimated both using oscilloscope and calorimeter. These results are found to be in good agreement with the simulated results.

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.

Differences between FRP bond behavior in cracked and uncracked regions

Based on an analysis of the experimental results of a proposed bond test method, significant differences are shown to exist between the local FRP bond stress-slip relationships in the uncracked anchorage regions and in the regions between cracks. The proposed method simulates the bond behavior between the flexural cracks and anchorage regions of a flexurally FRP-strengthened RC beam. The boundary conditions, including the presence of cracks and steel, are shown to have significant effects on the local bond stress-slip models. The results showed that, at the same force, the bond stresses in the regions between cracks were lower than in regions outside the cracks, so the debonding formed in the anchorage regions. The local bond stress-slip models in the anchorage regions can be obtained from the conventional bond test methods but these do not mimic the conditions between the cracks.