Effective handling of dialogue state in the hidden information state POMDP-based dialogue manager

Effective dialogue management is critically dependent on the information that is encoded in the dialogue state. In order to deploy reinforcement learning for policy optimization, dialogue must be modeled as a Markov Decision Process. This requires that the dialogue statemust encode all relevent information obtained during the dialogue prior to that state. This can be achieved by combining the user goal, the dialogue history, and the last user action to form the dialogue state. In addition, to gain robustness to input errors, dialogue must be modeled as a Partially Observable Markov Decision Process (POMDP) and hence, a distribution over all possible states must be maintained at every dialogue turn. This poses a potential computational limitation since there can be a very large number of dialogue states. The Hidden Information State model provides a principled way of ensuring tractability in a POMDP-based dialogue model. The key feature of this model is the grouping of user goals into partitions that are dynamically built during the dialogue. In this article, we extend this model further to incorporate the notion of complements. This allows for a more complex user goal to be represented, and it enables an effective pruning technique to be implemented that preserves the overall system performance within a limited computational resource more effectively than existing approaches. © 2011 ACM.

Differential amplitude modulation of auditory evoked cortical potentials associated with brain state in the freely moving rhesus monkey

We simultaneously recorded auditory evoked potentials (AEP) from the temporal cortex (TCx), the dorsolateral prefrontal cortex (dPFCx) and the parietal cortex (PCx) in the freely moving rhesus monkey to investigate state-dependent changes of the AEP. AEPs obtained during passive wakefulness, active wakefulness (AW), slow wave sleep and rapid-eye-movement sleep (REM) were compared. Results showed that AEP from all three cerebral areas were modulated by brain states. However, the amplitude of AEP from dPFCx and PCx significantly appeared greater attenuation than that from the TCx during AW and REM. These results indicate that the modulation of brain state on AEP from all three cerebral areas investigated is not uniform, which suggests that different cerebral areas have differential functional contributions during sleep-wake cycle. (C) 2002 Elsevier Science Ireland Ltd.. All rights reserved.

Optical spectra and exciton state in vertically stacked self-assembled quantum discs

We study the oscillator strengths of the optical transitions of the vertically stacked self-assembled InAs quantum discs. The oscillator strengths change evidently when the two quantum discs are far apart from each other. A vertically applied electric held affects the oscillator strengths severely, while the oscillator strengths change slowly as the radius of one disc increases. We also studied the excitonic energy of the system, including the Coulomb interaction. The excitonic energy increases with the increasing radius of one disc, but decreases as a vertically applied electric field increases.

Formation of an fcc phase through a bcc metastable state in crystallization of charged colloidal particles

By in situ monitoring structural changes with the reflection spectrometer during the colloidal crystallization, we present direct experimental evidence of liquid-bcc-fcc phase transition in crystallization of charged colloidal particles, as a manifestation of the Ostwald's step rule. In addition, the lifetime of the bcc metastable structure in this system decreases significantly with increasing particle volume fraction, offering a possible explanation for "exceptions" to the step rule.

STUDIES FOR THE EQUATION OF STATE IN THE ISOSPIN ASYMMETRICAL NUCLEAR INTERACTIONS

In order to determine the equation of state in the isospin asymmetrical nuclear interactions, we have found the observables for extracting the information of them within the isospin-dependent quantum molecular dynamics in recent years. The several sensitive probes for extracting the information of the in-medium nucleon-nucleon cross section and the symmetry potential have found; meanwhile, their mechanisms are investigated in more details. The main point in this paper gives the summary for above probes and their outlook in the future.

Resonant two-photon cross-section calculations for transitions to 4f5d state in Pr3+: YAG using density matrix theory

The density matrix resonant two-photon absorption (TPA) theory is applied to a rare-earth ion-doped laser crystal. TPA cross sections for transitions from the ground state to the first 4f5d state in Pr3+:YAG are calculated. The results indicate the density matrix TPA theory is attractive in studying TPA in laser crystals. (C) 2000 Elsevier Science B.V. All rights reserved.

Calculations of cross sections of resonant two-photon transitions to the second excited 4f5d state in Pr3+: Y3Al5O12 using density matrix theory

The density matrix resonant two-photon absorption (TPA) theory applicable to laser crystals doped with rare earth ions is described. Using this theory, resonant TPA cross sections for transitions from the ground state to the second excited state of the 4f5d configuration in cm(4)s Pr3+:Y3Al5O12 are calculated. The peak value of TPA cross section calculated is 2.75 x 10(-50) cm(4)s which is very close to the previous experimental value 4 x 10(-50) cm(4) s. The good agreement of calculated data with measured values demonstrates that the density matrix resonant TPA theory can predict resonant TPA intensity much better than the standard second-order perturbation TPA theory.

Small angle X-ray scattering study of structural parameters of the aggregated state in polyamide 1010+BMI

The structural parameters of the aggregated state in the polyamide PA1010 and N,N'-bismaleimide-4,4'-diphenyl methane (BMI) system were computed by means of the desmearing intensity from SAXS and using the concept of the distance distribution function. The results revealed that the parameters Q, I(0), l(c) and L decreased with the increase BMI component, whereas O-s increased. The particle dimension Z for different BMI contents was less than 13.2 nm, and the maximum value of the distance distribution function P(Z) was found to be in the range Z = 6.5-7.0 nm.