State-space inference and learning with Gaussian processes

State-space inference and learning with Gaussian processes (GPs) is an unsolved problem. We propose a new, general methodology for inference and learning in nonlinear state-space models that are described probabilistically by non-parametric GP models. We apply the expectation maximization algorithm to iterate between inference in the latent state-space and learning the parameters of the underlying GP dynamics model. Copyright 2010 by the authors.

Robust filtering and smoothing with gaussian processes

We propose a principled algorithm for robust Bayesian filtering and smoothing in nonlinear stochastic dynamic systems when both the transition function and the measurement function are described by non-parametric Gaussian process (GP) models. GPs are gaining increasing importance in signal processing, machine learning, robotics, and control for representing unknown system functions by posterior probability distributions. This modern way of system identification is more robust than finding point estimates of a parametric function representation. Our principled filtering/smoothing approach for GP dynamic systems is based on analytic moment matching in the context of the forward-backward algorithm. Our numerical evaluations demonstrate the robustness of the proposed approach in situations where other state-of-the-art Gaussian filters and smoothers can fail. © 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.

Mode division multiplexing over 2km of OM2 fibre using rotationally optimized mode excitation with fibre coupler demultiplexer

A Spatial Light Modulator and a non-specialized multimode coupler are used together to provide sufficient channel isolation and modal bandwidth for 2x12.5Gbps NRZ over 2km of standard graded-index multimode fibre without DSP. © 2012 IEEE.

A high performance oxygen storage material for chemical looping processes with CO2 capture

Chemical looping combustion (CLC) is a novel combustion technology that involves cyclic reduction and oxidation of oxygen storage materials to provide oxygen for the combustion of fuels to CO2 and H2O, whilst giving a pure stream of CO2 suitable for sequestration or utilisation. Here, we report a method for preparing of oxygen storage materials from layered double hydroxides (LDHs) precursors and demonstrate their applications in the CLC process. The LDHs precursor enables homogeneous mixing of elements at the molecular level, giving a high degree of dispersion and high-loading of active metal oxide in the support after calcination. Using a Cu-Al LDH precursor as a prototype, we demonstrate that rational design of oxygen storage materials by material chemistry significantly improved the reactivity and stability in the high temperature redox cycles. We discovered that the presence of sodium-containing species were effective in inhibiting the formation of copper aluminates (CuAl2O4 or CuAlO 2) and stabilising the copper phase in an amorphous support over multiple redox cycles. A representative nanostructured Cu-based oxygen storage material derived from the LDH precursor showed stable gaseous O2 release capacity (∼5 wt%), stable oxygen storage capacity (∼12 wt%), and stable reaction rates during reversible phase changes between CuO-Cu 2O-Cu at high temperatures (800-1000 °C). We anticipate that the strategy can be extended to manufacture a variety of metal oxide composites for applications in novel high temperature looping cycles for clean energy production and CO2 capture. © The Royal Society of Chemistry 2013.

New approaches for enhancing light emission from Er-based materials and devices

In this work, we present some approaches recently developed for enhancing light emission from Er-based materials and devices. We have investigated the luminescence quenching processes limiting quantum efficiency in light-emitting devices based on Si nanoclusters (Si nc) or Er-doped Si nc. It is found that carrier injection, while needed to excite Si nc or Er ions through electron-hole recombination, at the same time produces an efficient non-radiative Auger de-excitation with trapped carriers. A strong light confinement and enhancement of Er emission at 1.54 μm in planar silicon-on-insulator waveguides containing a thin layer (slot) of SiO2 with Er-doped Si nc at the center of the Si core has been obtained. By measuring the guided photoluminescence from the cleaved edge of the sample, we have observed a more than fivefold enhancement of emission for the transverse magnetic mode over the transverse electric one at room temperature. Slot waveguides have also been integrated with a photonic crystal (PhC), consisting of a triangular lattice of holes. An enhancement by more than two orders of magnitude of the Er near-normal emission is observed when the transition is in resonance with an appropriate mode of the PhC slab. Finally, in order to increase the concentration of excitable Er ions, a completely different approach, based on Er disilicate thin films, has been explored. Under proper annealing conditions crystalline and chemically stable Er2Si2O7 films are obtained; these films exhibit a strong luminescence at 1.54 μm owing to the efficient reduction of the defect density. © 2008 Elsevier B.V. All rights reserved.

Erbium doped materials for a Si-based microphotonics

We have investigated the role of the Si excess on the photoluminescence properties of Er doped substoichiometric SiOx layers. We demonstrate that the Si excess has two competing roles: when agglomerated to form Si nanoclusters (Si-nc) it enhances the Er excitation efficiency but it also introduces new non-radiative decay channels. When Er is excited through an energy transfer from Si-nc, the beneficial effect on the enhanced excitation efficiency prevails and the Er emission increases with increasing Si content. Nevertheless the maximum excited Er fraction is only of the order of percent. In order to increase the concentration of excited Er ions, a different approach based on Er silicate thin film has been explored. Under proper annealing conditions, an efficient luminescence at 1535 nm is found and all of the Er ions in the material is optically active. The possibility to efficiently excite Er ions also through electron-hole mediated processes is demonstrated in nanometer-scale Er-Si-O/Si multilayers. These data are presented and discussed.

Modeling natural sounds with modulation cascade processes

Natural sounds are structured on many time-scales. A typical segment of speech, for example, contains features that span four orders of magnitude: Sentences ($\sim1$s); phonemes ($\sim10$−$1$ s); glottal pulses ($\sim 10$−$2$s); and formants ($\sim 10$−$3$s). The auditory system uses information from each of these time-scales to solve complicated tasks such as auditory scene analysis [1]. One route toward understanding how auditory processing accomplishes this analysis is to build neuroscience-inspired algorithms which solve similar tasks and to compare the properties of these algorithms with properties of auditory processing. There is however a discord: Current machine-audition algorithms largely concentrate on the shorter time-scale structures in sounds, and the longer structures are ignored. The reason for this is two-fold. Firstly, it is a difficult technical problem to construct an algorithm that utilises both sorts of information. Secondly, it is computationally demanding to simultaneously process data both at high resolution (to extract short temporal information) and for long duration (to extract long temporal information). The contribution of this work is to develop a new statistical model for natural sounds that captures structure across a wide range of time-scales, and to provide efficient learning and inference algorithms. We demonstrate the success of this approach on a missing data task.

Modelling and control of nonlinear systems using Gaussian processes with partial model information

Gaussian processes are gaining increasing popularity among the control community, in particular for the modelling of discrete time state space systems. However, it has not been clear how to incorporate model information, in the form of known state relationships, when using a Gaussian process as a predictive model. An obvious example of known prior information is position and velocity related states. Incorporation of such information would be beneficial both computationally and for faster dynamics learning. This paper introduces a method of achieving this, yielding faster dynamics learning and a reduction in computational effort from O(Dn2) to O((D - F)n2) in the prediction stage for a system with D states, F known state relationships and n observations. The effectiveness of the method is demonstrated through its inclusion in the PILCO learning algorithm with application to the swing-up and balance of a torque-limited pendulum and the balancing of a robotic unicycle in simulation. © 2012 IEEE.

Gaussian Processes for Machine Learning

The code provided here originally demonstrated the main algorithms from Rasmussen and Williams: Gaussian Processes for Machine Learning. It has since grown to allow more likelihood functions, further inference methods and a flexible framework for specifying GPs.