Repository logo
 

Scholarly Works - Engineering - Information Engineering

Browse

Recent Submissions

Now showing 1 - 20 of 24
  • ItemOpen AccessAccepted version Peer-reviewed
    Future word contexts in neural network language models
    (IEEE, 2017) Chen, X; Liu, X; Ragni, A; Wang, Y; Gales, MJF; Gales, Mark [0000-0002-5311-8219]
    Recently, bidirectional recurrent network language models (bi-RNNLMs) have been shown to outperform standard, unidirectional, recurrent neural network language models (uni-RNNLMs) on a range of speech recognition tasks. This indicates that future word context information beyond the word history can be useful. However, bi-RNNLMs pose a number of challenges as they make use of the complete previous and future word context information. This impacts both training efficiency and their use within a lattice rescoring framework. In this paper these issues are addressed by proposing a novel neural network structure, succeeding word RNNLMs (su-RNNLMs). Instead of using a recurrent unit to capture the complete future word contexts, a feedforward unit is used to model a finite number of succeeding, future, words. This model can be trained much more efficiently than bi-RNNLMs and can also be used for lattice rescoring. Experimental results on a meeting transcription task (AMI) show the proposed model consistently outperformed uni-RNNLMs and yield only a slight degradation compared to bi-RNNLMs in N-best rescoring. Additionally, performance improvements can be obtained using lattice rescoring and subsequent confusion network decoding.
  • ItemOpen AccessAccepted version Peer-reviewed
    Predicting personal traits from facial images using convolutional neural networks augmented with facial landmark information
    (AAAI Press, 2016-07) Lewenberg, Y; Bachrach, Y; Shankar, S; Criminisi, A
    We consider the task of predicting various traits of a person given an image of their face. We estimate both objective traits, such as gender, ethnicity and hair-color; as well as subjective traits, such as the emotion a person expresses or whether he is humorous or attractive. For sizeable experimentation, we contribute a new Face Attributes Dataset (FAD), having roughly 200,000 attribute labels for the above traits, for over 10,000 facial images. Due to the recent surge of research on Deep Convolutional Neural Networks (CNNs), we begin by using a CNN architecture for estimating facial attributes and show that they indeed provide an impressive baseline performance. To further improve performance, we propose a novel approach that incorporates facial landmark information for input images as an additional channel, helping the CNN learn better attribute-specific features so that the landmarks across various training images hold correspondence. We empirically analyse the performance of our method, showing consistent improvement over the baseline across traits.
  • ItemOpen Access
    Control of aircraft in the terminal manoeuvring area using parallelised sequential Monte Carlo
    (American Institute of Aeronautics and Astronautics (AIAA), 2013-08) Eele, A; Maciejowski, J; Chau, T; Luk, W; Maciejowski, Jan [0000-0001-8281-8364]
    This paper reports on the use of a parallelised Model Predictive Control, Sequential Monte Carlo algorithm for solving the problem of conflict resolution and aircraft trajectory control in air traffic management specifically around the terminal manoeuvring area of an airport. The target problem is nonlinear, highly constrained, non-convex and uses a single decision-maker with multiple aircraft. The implementation includes a spatio-temporal wind model and rolling window simulations for realistic ongoing scenarios. The method is capable of handling arriving and departing aircraft simultaneously including some with very low fuel remaining. A novel flow field is proposed to smooth the approach trajectories for arriving aircraft and all trajectories are planned in three dimensions. Massive parallelisation of the algorithm allows solution speeds to approach those required for real-time use.
  • ItemOpen Access
    Field programmable gate array based predictive control system for spacecraft rendezvous in elliptical orbits
    (Wiley, 2015) Hartley, EN; Maciejowski, JM; Hartley, Edward [0000-0001-5491-229X]; Maciejowski, Jan [0000-0001-8281-8364]
    SummaryA field programmable gate array (FPGA) based model predictive controller for two phases of spacecraft rendezvous is presented. Linear time‐varying prediction models are used to accommodate elliptical orbits, and a variable prediction horizon is used to facilitate finite time completion of the longer range manoeuvres, whilst a fixed and receding prediction horizon is used for fine‐grained tracking at close range. The resulting constrained optimisation problems are solved using a primal–dual interior point algorithm. The majority of the computational demand is in solving a system of simultaneous linear equations at each iteration of this algorithm. To accelerate these operations, a custom circuit is implemented, using a combination of Mathworks HDL Coder and Xilinx System Generator for DSP, and used as a peripheral to a MicroBlaze soft‐core processor on the FPGA, on which the remainder of the system is implemented. Certain logic that can be hard‐coded for fixed sized problems is implemented to be configurable online, in order to accommodate the varying problem sizes associated with the variable prediction horizon. The system is demonstrated in closed‐loop by linking the FPGA with a simulation of the spacecraft dynamics running in Simulink on a PC, using Ethernet. Timing comparisons indicate that the custom implementation is substantially faster than pure embedded software‐based interior point methods running on the same MicroBlaze and could be competitive with a pure custom hardware implementation.Copyright © 2014 John Wiley & Sons, Ltd.
  • ItemOpen Access
    Comparison of stochastic methods for control in air traffic management
    (Elsevier BV, 2011-01) Eele, A; MacIejowski, J; Maciejowski, Jan [0000-0001-8281-8364]
    This paper provides a direct comparison of two stochastic optimisation techniques (Markov Chain Monte Carlo and Sequential Monte Carlo) when applied to the problem of conflict resolution and aircraft trajectory control in air traffic management. The two methods are then also compared to another existing technique of Mixed-Integer Linear Programming which is also popular in distributed control.
  • ItemOpen Access
    Parallelisation of sequential Monte Carlo for real-time control in air traffic management
    (IEEE, 2013-12) Eele, A; Maciejowski, J; Chau, T; Luk, W; Maciejowski, Jan [0000-0001-8281-8364]
    This paper presents the parallelisation of a Sequential Monte Carlo algorithm, and the associated changes required when applied to the problem of conflict resolution and aircraft trajectory control in air traffic management. The target problem is non-linear, constrained, non-convex and multi-agent. The new method is shown to have a 98.5% computational time saving over that of a previous sequential implementation, with no degradation in path quality. The computation saving is enough to allow real-time implementation.
  • ItemOpen Access
    Reverse engineered MPC for tracking with systems that become uncertain
    (IEEE, 2014) Hartley, EN; Maciejowski, JM; Hartley, Edward [0000-0001-5491-229X]; Maciejowski, Jan [0000-0001-8281-8364]
    A constrained model predictive control technique for tracking is proposed for systems whose models become uncertain (for example after a sensor failure). A linear time invariant robust controller with integral action is used as a baseline and ``reverse engineered'' into the form of a reduced order observer, steady state target calculator and control gain, based on a nominal model, augmented with integrating disturbance states. Constraints are enforced by optimising over perturbations to the nominal control action. Clean transition between a nominal, high performance mode of operation when parameters are known, to a safe and recursively feasible robust mode when parameters are unknown can be facilitated by using the same steady state target in both cases.
  • ItemOpen Access
    Predictive control using an FPGA with application to aircraft control
    (Institute of Electrical and Electronics Engineers (IEEE), 2014) Hartley, EN; Jerez, JL; Suardi, A; MacIejowski, JM; Kerrigan, EC; Constantinides, GA; Hartley, Edward [0000-0001-5491-229X]; Maciejowski, Jan [0000-0001-8281-8364]
    Alternative and more efficient computational methods can extend the applicability of MPC to systems with tight real-time requirements. This paper presents a ``system-on-a-chip'' MPC system, implemented on a field programmable gate array (FPGA), consisting of a sparse structure-exploiting primal dual interior point (PDIP) QP solver for MPC reference tracking and a fast gradient QP solver for steady-state target calculation. A parallel reduced precision iterative solver is used to accelerate the solution of the set of linear equations forming the computational bottleneck of the PDIP algorithm. A numerical study of the effect of reducing the number of iterations highlights the effectiveness of the approach. The system is demonstrated with an FPGA-in-the-loop testbench controlling a nonlinear simulation of a large airliner. This study considers many more manipulated inputs than any previous FPGA-based MPC implementation to date, yet the implementation comfortably fits into a mid-range FPGA, and the controller compares well in terms of solution quality and latency to state-of-the-art QP solvers running on a standard PC.
  • ItemOpen Access
    Designing output-feedback predictive controllers by reverse-engineering existing LTI controllers
    (Institute of Electrical and Electronics Engineers (IEEE), 2013) Hartley, EN; MacIejowski, JM; Hartley, Edward [0000-0001-5491-229X]; Maciejowski, Jan [0000-0001-8281-8364]
    An approach to designing a constrained output-feedback predictive controller that has the same small-signal properties as a pre-existing output-feedback linear time invariant controller is proposed. Systematic guidelines are proposed to select an appropriate (non-unique) realization of the resulting state observer. A method is proposed to transform a class of offset-free reference tracking controllers into the combination of an observer, steady-state target calculator and predictive controller. The procedure is demonstrated with a numerical example.
  • ItemOpen Access
    Terminal spacecraft rendezvous and capture with LASSO model predictive control
    (Informa UK Limited, 2013) Hartley, EN; Gallieri, M; Maciejowski, JM; Hartley, Edward [0000-0001-5491-229X]; Maciejowski, Jan [0000-0001-8281-8364]
    The recently investigated $\ell_{\mathrm{asso}}$ model predictive control (MPC) is applied to the terminal phase of a spacecraft rendezvous and capture mission. The interaction between the cost function and the treatment of minimum impulse bit (MIB) is also investigated. The propellant consumption with $\ell_{\mathrm{asso}}$ MPC for the considered scenario is noticeably less than with a conventional quadratic cost and control actions are sparser in time. Propellant consumption and sparsity are competitive with those achieved using a zone-based $\ell_1$ cost function, whilst requiring fewer decision variables in the optimisation problem than the latter. The $\ell_{\mathrm{asso}}$ MPC is demonstrated to meet tighter specifications on control precision, and also avoids the risk of undesirable behaviours often associated with pure $\ell_1$ stage costs.
  • ItemOpen Access
    Predictive control for spacecraft rendezvous in an elliptical orbit using an FPGA
    (IEEE, 2013) Hartley, EN; Maciejowski, JM; Hartley, Edward [0000-0001-5491-229X]; Maciejowski, Jan [0000-0001-8281-8364]
    A field programmable gate array (FPGA)-based predictive controller for a spacecraft rendezvous man{\oe}uvre is presented. A linear time varying prediction model is used to accommodate elliptical orbits, and a variable prediction horizon is used to facilitate finite time completion of man{\oe}uvres. The resulting constrained optimisation problems are solved using a primal dual interior point algorithm. The majority of the computational demand is in solving a set of linear equations at each iteration of this algorithm. To accelerate this operation, a custom circuit is implemented, using a combination of Mathworks HDL Coder and Xilinx System Generator for DSP, and used as a peripheral to a MicroBlaze soft core processor. The system is demonstrated in closed loop by linking the FPGA with a simulation of the plant dynamics running in Simulink on a PC, using Ethernet.
  • ItemOpen Access
    Graphical FPGA design for a predictive controller with application to spacecraft rendezvous
    (IEEE, 2013) Hartley, EN; Maciejowski, JM; Hartley, Edward [0000-0001-5491-229X]; Maciejowski, Jan [0000-0001-8281-8364]
    A reconfigurable field-programmable gate array (FPGA)-based predictive controller based on Nesterov’s fast gradient method is designed using Simulink and converted to VHDL using Mathworks’ HDL Coder. The implementation is verified by application to a spacecraft rendezvous and capture scenario, with communication between the FPGA and a simulation of the relative dynamics occuring over Ethernet. For a problem with 120 decision variables and 240 constraints, computation times of 0.95 ms are achieved with a clock rate of 50 MHz, corresponding to a speed up of more than 2000 over running the algorithm directly on a MicroBlaze microprocessor implemented on the same FPGA.
  • ItemOpen Access
    Performance evaluation of multiplexed model predictive control for a large airliner in nominal and contingency scenarios
    (IEEE, 2012) Hartley, EN; MacIejowski, JM; Ling, KV; Hartley, Edward [0000-0001-5491-229X]; Maciejowski, Jan [0000-0001-8281-8364]
    Model predictive control allows systematic han- dling of physical and operational constraints through the use of constrained optimisation. It has also been shown to successfully exploit plant redundancy to maintain a level of control in scenarios when faults are present. Unfortunately, the computa- tional complexity of each individual iteration of the algorithm to solve the optimisation problem scales cubically with the number of plant inputs, so the computational demands are high for large MIMO plants. Multiplexed MPC only calculates changes in a subset of the plant inputs at each sampling instant, thus reducing the complexity of the optimisation. This paper demonstrates the application of multiplexed model predictive control to a large transport airliner in a nominal and a contingency scenario. The performance is compared to that obtained with a conventional synchronous model predictive controller, designed using an equivalent cost function.
  • ItemOpen Access
    Predictive control of a Boeing 747 aircraft using an FPGA
    (Elsevier BV, 2012) Hartley, EN; Jerez, JL; Suardi, A; Maciejowski, JM; Kerrigan, EC; Constantinides, GA; Hartley, Edward [0000-0001-5491-229X]; Maciejowski, Jan [0000-0001-8281-8364]
    New embedded predictive control applications call for more efficient ways of solving quadratic programs (QPs) in order to meet demanding real-time, power and cost requirements. A single precision QP-on-a-chip controller is proposed, implemented in a field-programmable gate array (FPGA) with an iterative linear solver at its core. A novel offline scaling procedure is introduced to aid the convergence of the reduced precision solver. The feasibility of the proposed approach is demonstrated with a real-time hardware-in-the-loop (HIL) experimental setup where an ML605 FPGA board controls a nonlinear model of a Boeing 747 aircraft running on a desktop PC through an Ethernet link. Simulations show that the quality of the closed-loop control and accuracy of individual solutions is competitive with a conventional double precision controller solving linear systems using a Riccati recursion.
  • ItemOpen Access
    Initial tuning of predictive controllers by reverse engineering
    (IEEE, 2009) Hartley, EN; Maciejowski, JM; Hartley, Edward [0000-0001-5491-229X]; Maciejowski, Jan [0000-0001-8281-8364]
    This paper demonstrates a method for finding the cost function and state observer to be used in model predictive control (MPC) so that when constraints are inactive a pre- existing low order controller is reproduced. The MPC controller thereby inherits its desirable properties. This can be used as a baseline for further tuning. The available degrees of design freedom are explored, in order to facilitate, as appropriate, exploitation of constraint-handling, offset-free and redundancy management capabilities of MPC.
  • ItemOpen Access
    Fast, low-artifact speech synthesis considering global variance
    (IEEE (Institute of Electrical and Electronics Engineers), 2013-05-27) Shannon, Matt; Byrne, William
    Speech parameter generation considering global variance (GV generation) is widely acknowledged to dramatically improve the quality of synthetic speech generated by HMM-based systems. However it is slower and has higher latency than the standard speech parameter generation algorithm. In addition it is known to produce artifacts, though existing approaches to prevent artifacts are effective. We present a simple new theoretical analysis of speech parameter generation considering global variance based on Lagrange multipliers. This analysis sheds light on one source of artifacts and suggests a way to reduce their occurrence. It also suggests an approximation to exact GV generation that allows fast, low latency synthesis. In a subjective evaluation our fast approximation shows no degradation in naturalness compared to conventional GV generation.
  • ItemOpen Access
    Autoregressive Models for Statistical Parametric Speech Synthesis
    (Institute of Electrical and Electronics Engineers (IEEE), 2013-03) Shannon, M; Heiga Zen; Byrne, W
    We propose using the autoregressive hidden Markov model (HMM) for speech synthesis. The autoregressive HMM uses the same model for parameter estimation and synthesis in a consistent way, in contrast to the standard approach to statistical parametric speech synthesis. It supports easy and efficient parameter estimation using expectation maximization, in contrast to the trajectory HMM. At the same time its similarities to the standard approach allow use of established high quality synthesis algorithms such as speech parameter generation considering global variance. The autoregressive HMM also supports a speech parameter generation algorithm not available for the standard approach or the trajectory HMM and which has particular advantages in the domain of real-time, low latency synthesis. We show how to do efficient parameter estimation and synthesis with the autoregressive HMM and look at some of the similarities and differences between the standard approach, the trajectory HMM and the autoregressive HMM. We compare the three approaches in subjective and objective evaluations. We also systematically investigate which choices of parameters such as autoregressive order and number of states are optimal for the autoregressive HMM.
  • ItemOpen Access
    The effect of using normalized models in statistical speech synthesis
    (ISCA (International Speech Communication Association), 2011-08-27) Shannon, Matt; Zen, Heiga; Byrne, William
    The standard approach to HMM-based speech synthesis is inconsistent in the enforcement of the deterministic constraints between static and dynamic features. The trajectory HMM and autoregressive HMM have been proposed as normalized models which rectify this inconsistency. This paper investigates the practical effects of using these normalized models, and examines the strengths and weaknesses of the different models as probabilistic models of speech. The most striking difference observed is that the standard approach greatly underestimates predictive variance. We argue that the normalized models have better predictive distributions than the standard approach, but that all the models we consider are still far from satisfactory probabilistic models of speech. We also present evidence that better intra-frame correlation modelling goes some way towards improving existing normalized models.
  • ItemOpen Access
    A formulation of the autoregressive HMM for speech synthesis
    (Department of Engineering, University of Cambridge, 2009-08-31) Shannon, Matt; Byrne, William
    We present a formulation of the autoregressive HMM for speech synthesis and compare it to the standard HMM synthesis framework and the trajectory HMM. We give details of how to do efficient parameter estimation and synthesis with the autoregressive HMM and discuss consequences of the autoregressive HMM model. There are substantial similarities between the three models, which we explore. The advantages of the autoregressive HMM are that it uses the same model for parameter estimation and synthesis in a consistent way, in contrast to the standard HMM synthesis framework, and that it supports easy and efficient parameter estimation, in contrast to the trajectory HMM.
  • ItemOpen Access
    Autoregressive clustering for HMM speech synthesis
    (Curran Associates, Inc., 2011) Shannon, SM; Byrne, WJ
    The autoregressive HMM has been shown to provide efficient parameter estimation and high-quality synthesis, but in previous experiments decision trees derived from a non-autoregressive system were used. In this paper we investigate the use of autoregressive clustering for autoregressive HMM-based speech synthesis. We describe decision tree clustering for the autoregressive HMM and highlight differences to the standard clustering procedure. Subjective listening evaluation results suggest that autoregressive clustering improves the naturalness of the resulting speech. We find that the standard minimum description length (MDL) criterion for selecting model complexity is inappropriate for the autoregressive HMM. Investigating the effect of model complexity on naturalness, we find that a large degree of overfitting is tolerated without a substantial decrease in naturalness.