Uncertainty management for on-line optimisation of a POMDP-based large-scale spoken dialogue system

The optimization of dialogue policies using reinforcement learning (RL) is now an accepted part of the state of the art in spoken dialogue systems (SDS). Yet, it is still the case that the commonly used training algorithms for SDS require a large number of dialogues and hence most systems still rely on artificial data generated by a user simulator. Optimization is therefore performed off-line before releasing the system to real users. Gaussian Processes (GP) for RL have recently been applied to dialogue systems. One advantage of GP is that they compute an explicit measure of uncertainty in the value function estimates computed during learning. In this paper, a class of novel learning strategies is described which use uncertainty to control exploration on-line. Comparisons between several exploration schemes show that significant improvements to learning speed can be obtained and that rapid and safe online optimisation is possible, even on a complex task. Copyright © 2011 ISCA.

Anxiety reduction through detachment: subjective, physiological, and neural effects.

The ability to volitionally regulate emotions helps to adapt behavior to changing environmental demands and can alleviate subjective distress. We show that a cognitive strategy of detachment attenuates subjective and physiological measures of anticipatory anxiety for pain and reduces reactivity to receipt of pain itself. Using functional magnetic resonance imaging, we locate the potential site and source of this modulation of anticipatory anxiety in the medial prefrontal/anterior cingulate and anterolateral prefrontal cortex, respectively.

Towards robust design of axial compressors with uncertainty quantification

Operational uncertainties such as throttle excursions, varying inlet conditions and geometry changes lead to variability in compressor performance. In this work, the main operational uncertainties inherent in a transonic axial compressor are quantified to deter- mine their effect on performance. These uncertainties include the effects of inlet distortion, metal expansion, ow leakages and blade roughness. A 3D, validated RANS model of the compressor is utilized to simulate these uncertainties and quantify their effect on polytropic efficiency and pressure ratio. To propagate them, stochastic collocation and sparse pseudospectral approximations are used. We demonstrate that lower-order approximations are sufficient as these uncertainties are inherently linear. Results for epistemic uncertainties in the form of meshing methodologies are also presented. Finally, the uncertainties considered are ranked in order of their effect on efficiency loss. © 2012 AIAA.