Unsupervised Dialogue State Tracking for End-to-End Task-Oriented Dialogue with a Multi-Span Prediction Network

This paper focuses on end-to-end task-oriented dialogue systems,which jointly handle dialogue state tracking(DST)and response generation.Traditional methods usually adopt a supervised paradigm to learn DST from a manually labeled corpus.However,the annotation of the corpus is costly,time-consuming,and cannot cover a wide range of domains in the real world.To solve this problem,we propose a multi-span prediction network(MSPN)that performs unsupervised DST for end-to-end task-oriented dialogue.Specifically,MSPN contains a novel split-merge copy mechanism that captures long-term dependencies in dialogues to automatically extract multiple text spans as keywords.Based on these keywords,MSPN uses a semantic distance based clustering approach to obtain the values of each slot.In addition,we propose an ontology-based reinforcement learning approach,which employs the values of each slot to train MSPN to generate relevant values.Experimental results on single-domain and multi-domain task-oriented dialogue datasets show that MSPN achieves state-of-the-art performance with significant improvements.Besides,we construct a new Chinese dialogue dataset MeDial in the low-resource medical domain,which further demonstrates the adaptability of MSPN.

State tracking model reference adaptive control for switched nonlinear systems with linear uncertain parameters

Model reference adaptive control （MRAC） is considered for a class of switched nonlinear systems in which the unknown parameters appear linearly. The linear uncertain parameters in each subsystem can be expressed as a vector and the uncertain vectors in different subsystems are estimated individually by different vector variables. Update laws are designed such that the parameter estimation will ＇freeze＇ until its corresponding subsystem is active. Controllers for subsystems are given to ensure asymptotic states tracking under arbitrary switchings. Two examples are presented to validate the proposed method.

A Unified Shared-Private Network with Denoising for Dialogue State Tracking

Dialogue state tracking(DST)leverages dialogue information to predict dialogues states which are generally represented as slot-value pairs.However,previous work usually has limitations to efficiently predict values due to the lack of a powerful strategy for generating values from both the dialogue history and the predefined values.By predicting values from the predefined value set,previous discriminative DST methods are difficult to handle unknown values.Previous generative DST methods determine values based on mentions in the dialogue history,which makes it difficult for them to handle uncovered and non-pointable mentions.Besides,existing generative DST methods usually ignore the unlabeled instances and suffer from the label noise problem,which limits the generation of mentions and eventually hurts performance.In this paper,we propose a unified shared-private network(USPN)to generate values from both the dialogue history and the predefined values through a unified strategy.Specifically,USPN uses an encoder to construct a complete generative space for each slot and to discern shared information between slots through a shared-private architecture.Then,our model predicts values from the generative space through a shared-private decoder.We further utilize reinforcement learning to alleviate the label noise problem by learning indirect supervision from semantic relations between conversational words and predefined slot-value pairs.Experimental results on three public datasets show the effectiveness of USPN by outperforming state-of-the-art baselines in both supervised and unsupervised DST tasks.

Evolvable dialogue state tracking for statistical dialogue management

Statistical dialogue management is the core of cognitive spoken dialogue systems （SDS） and has attracted great research interest. In recent years, SDS with the ability of evolution is of particular interest and becomes the cuttingedge of SDS research. Dialogue state tracking （DST） is a process to estimate the distribution of the dialogue states at each dialogue turn, given the previous interaction history. It plays an important role in statistical dialogue management. To provide a common testbed for advancing the research of DST, international DST challenges （DSTC） have been organised and well-attended by major SDS groups in the world. This paper reviews recent progresses on rule-based and statistical approaches during the challenges. In particular, this paper is focused on evolvable DST approaches for dialogue domain extension. The two primary aspects for evolution, semantic parsing and tracker, are discussed. Semantic enhancement and a DST framework which bridges rule-based and statistical models are introduced in detail. By effectively incorporating prior knowledge of dialogue state transition and the ability of being data-driven, the new framework supports reliable domain extension with little data and can continuously improve with more data available. This makes it excellent candidate for DST evolution. Experiments show that the evolvable DST approaches can achieve the state-of-the-art performance and outperform all previously submitted trackers in the third DSTC.

Exploiting More Associations Between Slots for Multi-Domain Dialog State Tracking

Dialog State Tracking(DST)aims to extract the current state from the conversation and plays an important role in dialog systems.Existing methods usually predict the value of each slot independently and do not consider the correlations among slots,which will exacerbate the data sparsity problem because of the increased number of candidate values.In this paper,we propose a multi-domain DST model that integrates slot-relevant information.In particular,certain connections may exist among slots in different domains,and their corresponding values can be obtained through explicit or implicit reasoning.Therefore,we use the graph adjacency matrix to determine the correlation between slots,so that the slots can incorporate more slot-value transformer information.Experimental results show that our approach has performed well on the Multi-domain Wizard-Of-Oz(MultiWOZ)2.0and MultiWOZ2.1 datasets,demonstrating the effectiveness and necessity of incorporating slot-relevant information.

MULTITARGET STATE AND TRACK ESTIMATION FOR THE PROBABILITY HYPOTHESES DENSITY FILTER

The particle Probability Hypotheses Density (particle-PHD) filter is a tractable approach for Random Finite Set (RFS) Bayes estimation, but the particle-PHD filter can not directly derive the target track. Most existing approaches combine the data association step to solve this problem. This paper proposes an algorithm which does not need the association step. Our basic ideal is based on the clustering algorithm of Finite Mixture Models (FMM). The intensity distribution is first derived by the particle-PHD filter, and then the clustering algorithm is applied to estimate the multitarget states and tracks jointly. The clustering process includes two steps: the prediction and update. The key to the proposed algorithm is to use the prediction as the initial points and the convergent points as the es- timates. Besides, Expectation-Maximization (EM) and Markov Chain Monte Carlo (MCMC) ap- proaches are used for the FMM parameter estimation.

Tracking Power System State Evolution with Maximum-correntropy-based Extended Kalman Filter

This paper develops a novel approach to track power system state evolution based on the maximum correntropy criterion,due to its robustness against non-Gaussian errors.It includes the temporal aspects on the estimation process within a maximum-correntropy-based extended Kalman filter(MCEKF),which is able to deal with both nonlinear supervisory control and data acquisition(SCADA)and phasor measurement unit(PMU)measurement models.By representing the behavior of the state variables with a nonparametric model within the kernel density estimation,it is possible to include abrupt state transitions as part of the process noise with non-Gaussian characteristics.Also,a novel strategy to update the size of Parzen windows in the kernel estimation is proposed to suppress the effects of suspect samples.By properly adjusting the kernel bandwidth,the proposed MCEKF keeps its accuracy during sudden load changes and contingencies,or in the presence of bad data.Simulations with IEEE test systems and the Brazilian interconnected system are carried out.The results show that the method deals with non-Gaussian noises in both the process and measurement,and provides accurate estimates of the system state under normal and abnormal conditions.

Best linear unbiased estimation algorithm with Doppler measurements in spherical coordinates

In an active radar-tracking system,the target-motion model is usually modeled in the Cartesian coordinates,while the radar measurement usually is obtained in polar/spherical coordinates.Therefore the target-tracking problem in the Cartesian coordinates becomes a nonlinear state estimation problem.A number of measurement-conversion techniques,which are based on position measurements,are widely used such that the Kalman filter can be used in the Cartesian coordinates.However,they have fundamental limitations to result in filtering performance degradation.In fact,in addition to position measurements,the Doppler measurement or range rate,containing information of target velocity,has the potential capability to improve the tracking performance.A filter is proposed that can use converted Doppler measurements（i.e.the product of the range measurements and Doppler measurements） in the Cartesian coordinates.The novel filter is theoretically optimal in the rule of the best linear unbiased estimation among all linear unbiased filters in the Cartesian coordinates,and is free of the fundamental limitations of the measurement-conversion approach.Based on simulation experiments,an approximate,recursive implementation of the novel filter is compared with those obtained by four state-of-the-art conversion techniques recently.Simulation results demonstrate the effectiveness of the proposed filter.

Indoor^(222)Rn Levels and Effective Dose Estimation of Academic Staff in izmir-Turkey

Objective To investigate the annual effective doses from indoor radon received by academic staff in the Faculty building. Methods Measurements of indoor radon concentrations were performed in the Arts and Sciences Faculty of Dokuz Eylul University for two surveys of about 1 month duration respectively using the SSNTD （Solid State Nuclear Track Detectors） method with LR115 detectors. Time integrated measurements comprised different locations inside the faculty building： classrooms, toilets, canteen and offices. Homes of academic staff were also tested for radon. Results The aritthmetic mean radon concentration is 161 Bq m-3 with a range between 40 and 335 Bq m-3 in the Faculty. Six offices and three classrooms have a radon concentration above 200 Bq m-3. The results show that the radon concentration in classrooms is generally higher than in offices. Based on the measured indoor radon data, the annual effective doses received by staff in the Faculty were estimated to range from 0.79 to 4.27 mSv, according to UNSCEAR methodology. The annual effective doses received by staff ranged from 0.78 to 4.20 mSv in homes. On average, the Faculty contributed 56% to the annual effective dose. Conclusion Reported values for radon concentrations and corresponding doses are within the ICRP recommended limits for workplaces.

Neutron activation analysis of ceramic tiles and its component and radon exhalation rate

The concentrations of 20 trace elements in several ceramics tiles and ceramic composites used in Egypt were elementally analyzed by neutron activation analysis(NAA) technique. The samples and standard were irradiated with reactor for 4 h(in the Second Research Egyptian Reactor(Et-RR-2)) with thermal neutron flux 5.9×10 13 n/(cm 2·s).The gamma-ray spectra obtained were measured for several times by means of the hyper pure germanium detection system(HPGe). Also a solid state nuclear track detector(SSNTD) CR-39, was used to measure the emanation rate of radon for these samples. The radium concentrations were found to vary from 0.39—3.59 ppm and the emanation rates were found to vary from (0.728—5.688) × 10 -4 kg/(m 2· s).The elemental analysis of the ceramic tiles and ceramic composites have a great importance in assigning the physical properties and in turn the quality of the material.

Characteristics of Ions Emitted from Laser-Induced Silver Plasma

In this work, study of laser-induced ions is presented. The plasma was produced by focusing a Nd：YAG laser, with a wavelength of 1064 nm, a pulsed width of 9-14 ns, a power of 1.1 MW and energy of 10 mJ, on silver target in vacuum （10-3 Torr= 1.3332 Pa）. The characteristics of ion streams were investigated by CR-39 detectors located at angles of 0°, 30°, 60° and 90° with respect to normal of the target. The distance between the silver target and each detector was 11 cm. The energy of silver ions was found ranging from 1.5 eV to 1.06E4 eV. There was a high concentration of ions with low energy as compared to those with high energy, showing the energy distribution amongst the ions. The flux of ions was maximum in the axial direction which was decreasing with the angle increase with respect to normal of the target, and finally became minimum in the radial direction. Hence the silver ions have shown anisotropic behaviour.

Measurement of Radon Gas Concentrations in Tap Water Samples for Thi-Qar Governorate Using Nuclear Track Detector (CR-39)

In the present work, we have measured the radon gas concentrations in tap water samples are taken directly from drinking tap water in sites houses being carried in Thi-Qar governorate by using nuclear track detector (CR-39). The results of measurements have shown that the highest average radon concentration in water samples is found in AL-Refai region which is equal to (0.223 ± 0.03 Bq/L), while the lowest average radon gas concentration is found in AL-Fajr region which is equal to (0.108 ± 0.01 Bq/L), with an average value of (0.175 ± 0.03 Bq/L). The highest value of annual effective dose (AED) in tap water samples is found in AL-Refai region, which is equal to (0.814 μSv/y), while the lowest value of (AED) is found in AL-Fajr region which is equal to (0.394 μSv/y), with an average value of (0.640 ± 0.1 μSv/y). The present results have shown that radon gas concentrations in tap water samples are less than the recommended international value (11.1 Bq/L). There for tap water in all the studied sites in Thi-Qar governorate is safe as for as radon concentration being concerned.

Recent advances and challenges in task-oriented dialog systems

Due to the significance and value in human-computer interaction and natural language processing,task-oriented dialog systems are attracting more and more attention in both academic and industrial communities.In this paper,we survey recent advances and challenges in task-oriented dialog systems.We also discuss three critical topics for task-oriented dialog systems:(1)improving data efficiency to facilitate dialog modeling in low-resource settings,(2)modeling multi-turn dynamics for dialog policy learning to achieve better task-completion performance,and(3)integrating domain ontology knowledge into the dialog model.Besides,we review the recent progresses in dialog evaluation and some widely-used corpora.We believe that this survey,though incomplete,can shed a light on future research in task-oriented dialog systems.

Research Program of Determination of Geological Age by Thermal Analysis Method

On the basis of previous research achievements of measuring the solid state nuclear track in apatite by thermal analysis method, the author further proposes the research program to measure the energy deposited by the solid state nuclear track contained in zircon, sphene, epidote, apatite and other samples, in order to study the geological age and geothermal history. Compared with the measurement of nuclear track density by etching method, this one does not need to conduct so many processing programs for samples, but can improve the measurement accuracy.