The design and implementation of dialogue control in voice operated database inquiry systems

This paper describes work performed as part of the U.K. Alvey sponsored Voice Operated Database Inquiry System (VODIS) project in the area of intelligent dialogue control. The principal aims of the work were to develop a habitable interface for the untrained user; to investigate the degree to which dialogue control can be used to compensate for deficiencies in recognition performance; and to examine the requirements on dialogue control for generating natural speech output. A data-driven methodology is described based on the use of frames in which dialogue topics are organized hierarchically. The concept of a dynamically adjustable scope is introduced to permit adaptation to recognizer performance and the use of historical and hierarchical contexts are described to facilitate the construction of contextually relevant output messages. © 1989.

Multi-objective design and optimisation of bypass outlet-guide vanes

This paper describes the development of an automated design optimization system that makes use of a high fidelity Reynolds-Averaged CFD analysis procedure to minimize the fan forcing and fan BOGV (bypass outlet guide vane) losses simultaneously taking into the account the down-stream pylon and RDF (radial drive fairing) distortions. The design space consists of the OGV's stagger angle, trailing-edge recambering, axial and circumferential positions leading to a variable pitch optimum design. An advanced optimization system called SOFT (Smart Optimisation for Turbomachinery) was used to integrate a number of pre-processor, simulation and in-house grid generation codes and postprocessor programs. A number of multi-objective, multi-point optimiztion were carried out by SOFT on a cluster of workstations and are reported herein.

Analysis of in-cylinder hydrocarbons in a multi-cylinder gasoline HCCI engine using gas chromatography

Gasoline Homogeneous Charge Compression Ignition (HCCI) combustion has been studied widely in the past decade. However, in HCCI engines using negative valve overlap (NVO), there is still uncertainty as to whether the effect of pilot injection during NVO on the start of combustion is primarily due to heat release of the pilot fuel during NVO or whether it is due to pilot fuel reformation. This paper presents data taken on a 4-cylinder gasoline direct injection, spark ignition/HCCI engine with a dual cam system, capable of recompressing residual gas. Engine in-cylinder samples are extracted at various points during the engine cycle through a high-speed sampling system and directly analysed with a gas chromatograph and flame ionisation detector. Engine parameter sweeps are performed for different pilot injection timings and quantities at a medium load point. Results show that for lean engine running conditions, earlier pilot injection timing leads to partial oxidation of the injected pilot fuel during NVO, while the fraction of light hydrocarbons remains constant for all parameter variations investigated. The same applies for a variation in pilot fuel amount. Thus there is evidence that in lean conditions, pilot injection-related NVO effects are dominated by heat release rather than fuel reformation. © 2009 SAE International.

3D multi-car tracking based on monocular video

We propose a system that can reliably track multiple cars in congested traffic environments. Our system's key basis is the implementation of a sequential Monte Carlo algorithm, which introduces robustness against problems arising due to the proximity between vehicles. By directly modelling occlusions and collisions between cars we obtain promising results on an urban traffic dataset. Extensions to this initial framework are also suggested. © 2010 IEEE.

Corpus design for a unit selection TtS system with application to Bulgarian

In this paper we present the process of designing an efficient speech corpus for the first unit selection speech synthesis system for Bulgarian, along with some significant preliminary results regarding the quality of the resulted system. As the initial corpus is a crucial factor for the quality delivered by the Text-to-Speech system, special effort has been given in designing a complete and efficient corpus for use in a unit selection TTS system. The targeted domain of the TTS system and hence that of the corpus is the news reports, and although it is a restricted one, it is characterized by an unlimited vocabulary. The paper focuses on issues regarding the design of an optimal corpus for such a framework and the ideas on which our approach was based on. A novel multi-stage approach is presented, with special attention given to language and speaker dependent issues, as they affect the entire process. The paper concludes with the presentation of our results and the evaluation experiments, which provide clear evidence of the quality level achieved. © 2011 Springer-Verlag.

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.

Multi-objective aerodynamic design optimisation for axial compressors

This paper considers the aerodynamic design optimisation of turbomachinery blades from a multi-objective perspective. The aim is to improve the performance of a specific stage and eventually of the whole engine. The integrated system developed for this purpose is described. It combines an existing geometry parameterisation scheme, a well-established CFD package and a novel multi-objective variant of the Tabu Search optimisation algorithm. Its performance is illustrated through a case study in which the flow characteristics most important to the overall performance of turbomachinery blades are optimised.

An advanced multi-configuration stator well cooling test facility

Optimisation of cooling systems within gas turbine engines is of great interest to engine manufacturers seeking gains in performance, efficiency and component life. The effectiveness of coolant delivery is governed by complex flows within the stator wells and the interaction of main annulus and cooling air in the vicinity of the rim seals. This paper reports the development of a test facility which allows the interaction of cooling air and main gas paths to be measured at conditions representative of those found in modern gas turbine engines. The test facility features a two stage turbine with an overall pressure ratio of approximately 2.6:1. Hot air is supplied to the main annulus using a Rolls-Royce Dart compressor driven by an aero-derivative engine plant. Cooling air can be delivered to the stator wells at multiple locations and at a range of flow rates which cover bulk ingestion through to bulk egress. The facility has been designed with adaptable geometry to enable rapid changes of cooling air path configuration. The coolant delivery system allows swift and accurate changes to the flow settings such that thermal transients may be performed. Particular attention has been focused on obtaining high accuracy data, using a radio telemetry system, as well as thorough through-calibration practices. Temperature measurements can now be made on both rotating and stationary discs with a long term uncertainty in the region of 0.3 K. A gas concentration measurement system has also been developed to obtain direct measurement of re-ingestion and rim seal exchange flows. High resolution displacement sensors have been installed in order to measure hot running geometry. This paper documents the commissioning of a test facility which is unique in terms of rapid configuration changes, non-dimensional engine matching and the instrumentation density and resolution. Example data for each of the measurement systems is presented. This includes the effect of coolant flow rate on the metal temperatures within the upstream cavity of the turbine stator well, the axial displacement of the rotor assembly during a commissioning test, and the effect of coolant flow rate on mixing in the downstream cavity of the stator well. Copyright © 2010 by ASME.

Modal liquid crystal devices in optical tweezing: 3D control and oscillating potential wells

We investigate the use of liquid crystal (LC) adaptive optics elements to provide full 3 dimensional particle control in an optical tweezer. These devices are suitable for single controllable traps, and so are less versatile than many of the competing technologies which can be used to control multiple particles. However, they have the advantages of simplicity and light efficiency. Furthermore, compared to binary holographic optical traps they have increased positional accuracy. The transmissive LC devices could be retro-fitted to an existing microscope system. An adaptive modal LC lens is used to vary the z-focal position over a range of up to 100 μm and an adaptive LC beam-steering device is used to deflect the beam (and trapped particle) in the x-y plane within an available radius of 10 μm. Furthermore, by modifying the polarisation of the incident light, these LC components also offer the opportunity for the creation of dual optical traps of controllable depth and separation. © 2006 Optical Society of America.

Study of spike and modal stall phenomena in a low-speed axial compressor

This paper presents a study of stall inception mechanisms a in low-speed axial compressor. Previous work has identified two common flow breakdown sequences, the first associated with a short lengthscale disturbance known as a `spike', and the second with a longer lengthscale disturbance known as a `modal oscillation'. In this paper the physical differences between these two mechanisms are illustrated with detailed measurements. Experimental results are also presented which relate the occurrence of the two stalling mechanisms to the operating conditions of the compressor. It is shown that the stability criteria for the two disturbances are different: long lengthscale disturbances are related to a two-dimensional instability of the whole compression system, while short lengthscale disturbances indicate a three-dimensional breakdown of the flow-field associated with high rotor incidence angles. Based on the experimental measurements, a simple model is proposed which explains the type of stall inception pattern observed in a particular compressor. Measurements from a single stage low-speed compressor and from a multistage high-speed compressor are presented in support of the model.

Biologically-inspired object recognition system with features from complex wavelets

In this paper, a novel cortex-inspired feed-forward hierarchical object recognition system based on complex wavelets is proposed and tested. Complex wavelets contain three key properties for object representation: shift invariance, which enables the extraction of stable local features; good directional selectivity, which simplifies the determination of image orientations; and limited redundancy, which allows for efficient signal analysis using the multi-resolution decomposition offered by complex wavelets. In this paper, we propose a complete cortex-inspired object recognition system based on complex wavelets. We find that the implementation of the HMAX model for object recognition in [1, 2] is rather over-complete and includes too much redundant information and processing. We have optimized the structure of the model to make it more efficient. Specifically, we have used the Caltech 5 standard dataset to compare with Serre's model in [2] (which employs Gabor filter bands). Results demonstrate that the complex wavelet model achieves a speed improvement of about 4 times over the Serre model and gives comparable recognition performance. © 2011 IEEE.

Film bulk acoustic resonator nanosensors for multi-task sensing

This paper shows that film bulk acoustic resonator (FBAR) arrays can be very useful sensors either to detect physical parameters such as temperature and pressure directly or to detect bio-chemicals with extremely high sensitivities by incorporating a chemisorption layer or bio-probe molecules. Furthermore, it also shows that surface acoustic wave devices can be integrated with a FBAR sensor array on the same piezoelectric substrate as the microfluidics systems to perform transportation and mixing of biosamples etc. demonstrating the possibility to fabricate integrated lab-on-a-chip detection systems, in which all the actuators and sensors are operated by acoustic wave devices. This makes the detection system simple, low cost and easy to operate and hence has great commercial potential. © 2011 Inderscience Enterprises Ltd.

On-line policy optimisation of spoken dialogue systems via live interaction with human subjects

Statistical dialogue models have required a large number of dialogues to optimise the dialogue policy, relying on the use of a simulated user. This results in a mismatch between training and live conditions, and significant development costs for the simulator thereby mitigating many of the claimed benefits of such models. Recent work on Gaussian process reinforcement learning, has shown that learning can be substantially accelerated. This paper reports on an experiment to learn a policy for a real-world task directly from human interaction using rewards provided by users. It shows that a usable policy can be learnt in just a few hundred dialogues without needing a user simulator and, using a learning strategy that reduces the risk of taking bad actions. The paper also investigates adaptation behaviour when the system continues learning for several thousand dialogues and highlights the need for robustness to noisy rewards. © 2011 IEEE.

Partially Observable Markov Decision Processes with continuous observations for dialogue management

This work shows how a dialogue model can be represented as a Partially Observable Markov Decision Process (POMDP) with observations composed of a discrete and continuous component. The continuous component enables the model to directly incorporate a confidence score for automated planning. Using a testbed simulated dialogue management problem, we show how recent optimization techniques are able to find a policy for this continuous POMDP which outperforms a traditional MDP approach. Further, we present a method for automatically improving handcrafted dialogue managers by incorporating POMDP belief state monitoring, including confidence score information. Experiments on the testbed system show significant improvements for several example handcrafted dialogue managers across a range of operating conditions.

Improving LVCSR system combination using neural network language model cross adaptation

State-of-the-art large vocabulary continuous speech recognition (LVCSR) systems often combine outputs from multiple subsystems developed at different sites. Cross system adaptation can be used as an alternative to direct hypothesis level combination schemes such as ROVER. The standard approach involves only cross adapting acoustic models. To fully exploit the complimentary features among sub-systems, language model (LM) cross adaptation techniques can be used. Previous research on multi-level n-gram LM cross adaptation is extended to further include the cross adaptation of neural network LMs in this paper. Using this improved LM cross adaptation framework, significant error rate gains of 4.0%-7.1% relative were obtained over acoustic model only cross adaptation when combining a range of Chinese LVCSR sub-systems used in the 2010 and 2011 DARPA GALE evaluations. Copyright © 2011 ISCA.