Gaussian processes for POMDP-based dialogue manager optimization

A partially observable Markov decision process (POMDP) has been proposed as a dialog model that enables automatic optimization of the dialog policy and provides robustness to speech understanding errors. Various approximations allow such a model to be used for building real-world dialog systems. However, they require a large number of dialogs to train the dialog policy and hence they typically rely on the availability of a user simulator. They also require significant designer effort to hand-craft the policy representation. We investigate the use of Gaussian processes (GPs) in policy modeling to overcome these problems. We show that GP policy optimization can be implemented for a real world POMDP dialog manager, and in particular: 1) we examine different formulations of a GP policy to minimize variability in the learning process; 2) we find that the use of GP increases the learning rate by an order of magnitude thereby allowing learning by direct interaction with human users; and 3) we demonstrate that designer effort can be substantially reduced by basing the policy directly on the full belief space thereby avoiding ad hoc feature space modeling. Overall, the GP approach represents an important step forward towards fully automatic dialog policy optimization in real world systems. © 2013 IEEE.

20 Gb/s mode-group-division multiplexing employing Hermite-Gaussian launches over worst-case multimode fiber links

For the first time, mode group division multiplexing is achieved in a multimode fiber link using a 2-D Hermite-Gaussian mode launch. 20 Gb/s error-free transmission is achieved over a 250 m worst-case OM1 multimode fiber link. © OSA 2014.

Bayesian inference and learning in Gaussian process state-space models with Particle MCMC

State-space models are successfully used in many areas of science, engineering and economics to model time series and dynamical systems. We present a fully Bayesian approach to inference and learning (i.e. state estimation and system identification) in nonlinear nonparametric state-space models. We place a Gaussian process prior over the state transition dynamics, resulting in a flexible model able to capture complex dynamical phenomena. To enable efficient inference, we marginalize over the transition dynamics function and, instead, infer directly the joint smoothing distribution using specially tailored Particle Markov Chain Monte Carlo samplers. Once a sample from the smoothing distribution is computed, the state transition predictive distribution can be formulated analytically. Our approach preserves the full nonparametric expressivity of the model and can make use of sparse Gaussian processes to greatly reduce computational complexity.

Deterministic cold cathode electron emission from carbon nanofibre arrays.

The ability to accurately design carbon nanofibre (CN) field emitters with predictable electron emission characteristics will enable their use as electron sources in various applications such as microwave amplifiers, electron microscopy, parallel beam electron lithography and advanced Xray sources. Here, highly uniform CN arrays of controlled diameter, pitch and length were fabricated using plasma enhanced chemical vapour deposition and their individual emission characteristics and field enhancement factors were probed using scanning anode field emission mapping. For a pitch of 10 µm and a CN length of 5 µm, the directly measured enhancement factors of individual CNs was 242, which was in excellent agreement with conventional geometry estimates (240). We show here direct empirical evidence that in regular arrays of vertically aligned CNs the overall enhancement factor is reduced when the pitch between emitters is less than half the emitter height, in accordance to our electrostatic simulations. Individual emitters showed narrow Gaussian-like field enhancement distributions, in excellent agreement with electric field simulations.

Free space adaptive optical interconnect, using a ferroelectric liquid crystal SLM for beam steering

A free-space, board-to-board, adaptive optical interconnect demonstrator has been developed. Binary phase gratings displayed on a Ferroelectric Liquid Crystal Spatial Light Modulator are used to maintain data transfer at 1.25Gbps, given varying optical misalignment.© 2005 Optical Society of America.

Classification using log Gaussian Cox processes

McCullagh and Yang (2006) suggest a family of classification algorithms based on Cox processes. We further investigate the log Gaussian variant which has a number of appealing properties. Conditioned on the covariates, the distribution over labels is given by a type of conditional Markov random field. In the supervised case, computation of the predictive probability of a single test point scales linearly with the number of training points and the multiclass generalization is straightforward. We show new links between the supervised method and classical nonparametric methods. We give a detailed analysis of the pairwise graph representable Markov random field, which we use to extend the model to semi-supervised learning problems, and propose an inference method based on graph min-cuts. We give the first experimental analysis on supervised and semi-supervised datasets and show good empirical performance.

An energy-efficient hopping robot based on free vibration of a curved beam

This paper explores a design strategy of hopping robots, which makes use of free vibration of an elastic curved beam. In this strategy, the leg structure consists of a specifically shaped elastic curved beam and a small rotating mass that induces free vibration of the entire robot body. Although we expect to improve energy efficiency of locomotion by exploiting the mechanical dynamics, it is not trivial to take advantage of the coupled dynamics between actuation and mechanical structures for the purpose of locomotion. From this perspective, this paper explains the basic design principles through modeling, simulation, and experiments of a minimalistic hopping robot platform. More specifically, we show how to design elastic curved beams for stable hopping locomotion and the control method by using unconventional actuation. In addition, we also analyze the proposed design strategy in terms of energy efficiency and discuss how it can be applied to the other forms of legged robot locomotion. © 1996-2012 IEEE.

Off-axis sonar beam pattern of free-ranging finless porpoises measured by a stereo pulse event data logger

The off-axis sonar beam patterns of eight free-ranging finless porpoises were measured using attached data logger systems. The transmitted sound pressure level at each beam angle was calculated from the animal's body angle, the water surface echo level, and the swimming depth. The beam pattern of the off-axis signals between 45 and 115 (where 0 corresponds to the on-axis direction) was nearly constant. The sound pressure level of the off-axis signals reached 162 dB re 1 mPa peak-to-peak. The surface echo level received at the animal was over 140 dB, much higher than the auditory threshold level of small odontocetes. Finless porpoises are estimated to be able to receive the surface echoes of off-axis signals even at 50-m depth. Shallow water systems (less than 50-m depth) are the dominant habitat of both oceanic and freshwater populations of this species. Surface echoes may provide porpoises not only with diving depth information but also with information about surface direction and location of obstacles (including prey items) outside the on-axis sector of the sonar beam. 2005 Acoustical Society of America.

Hopping robot based on free vibration of an elastic curved beam

This study presents a novel approach to the design of low-cost and energy-efficient hopping robots, which makes use of free vibration of an elastic curved beam. We found that a hopping robot could benefit from an elastic curved beam in many ways such as low manufacturing cost, light body weight and small energy dissipation in mechanical interactions. A challenging problem of this design strategy, however, lies in harnessing the mechanical dynamics of free vibration in the elastic curved beam: because the free vibration is the outcome of coupled mechanical dynamics between actuation and mechanical structures, it is not trivial to systematically design mechanical structures and control architectures for stable locomotion. From this perspective, this paper investigates a case study of simple hopping robot to identify the design principles of mechanics and control. We developed a hopping robot consisting of an elastic curved beam and a small rotating mass, which was then modeled and analyzed in simulation. The experimental results show that the robot is capable of exhibiting stable hopping gait patterns by using a small actuation with no sensory feedback owing to the intrinsic stability of coupled mechanical dynamics. Furthermore, an additional analysis shows that, by exploiting free vibration of the elastic curved beam, cost of transport of the proposed hopping locomotion can be in the same rage of animals' locomotion including human running. © 2011 IEEE.

Fabrication of a Low-Loss SSC Using High-Dose Electron Beam Lithography Exposure With Negative PMMA Resist

A silicon-on-insulator optical fiber-to-waveguide spot-size converter (SSC) using Poly-MethylMethAcrylate (PMMA) is presented for integrated optical circuits. Unlike the conventional use of PMMA as a positive resist, it has been successfully used as a negative resist with high-dose electron exposure for the fabrication of ultrafine silicon wire waveguides. Additionally, this process is able to reduce the side-wall roughness, and substantially depresses the unwanted propagation loss. Exploiting this technology, the authors demonstrated that the SSC can improve coupling efficiency by as much as over 2.5 dB per coupling facet, compared with that of SSC fabricated with PMMA as a positive resist with the same dimension.

Design of a compact polarizing beam splitter based on a photonic crystal ring resonator with a triangular lattice

We propose and simulate a new kind of compact polarizing beam splitter (PBS) based on a photonic crystal ring resonator (PCRR) with complete photonic bandgaps. The two polarized states are separated far enough by resonant and nonresonant coupling between the waveguide modes and the microring modes. Some defect holes are utilized to control the beam propagation. The simulated results obtained by the finite-difference time-domain method show that high transmission (over 95%) is obtained and the polarization separation is realized with a length as short as 3.1 mu m. The design of the proposed PBS can be flexible, thanks to the advantages of PCRRs.

Design of novel polarization beam splitter in two-dimensional photonic crystal

We present the design and the simulation of an ultracompact high efficiency polarization beam splitter (PBS) based on the properties of the light waves propagating in straight waveguide and composite structure photonic crystal. The splitting properties of the PBS are numerically simulated and analyzed by using the plane wave expansion (PWE) method and finite difference time domain (FDTD) method. The PBS consists of three parts, namely, input waveguide, beam structure and output waveguide. It is shown that a high efficiency and a large separating angle for TE mode and TM mode can be achieved. Owing to these excellent features, including small size and high rate, the PBS makes a promising candidate in the future photonic integrated circuits.

Influence of AlN Buffer Thickness on GaN Grown on Si(111) by Gas Source Molecular Beam Epitaxy with Ammonia

Hexagonal GaN is grown on a Si(111) substrate with AlN as a buffer layer by gas source molecular beam epitaxy (GSMBE) with ammonia. The thickness of AlN buffer is changed from 9 to 72 nm. When the thickness of AlN buffer is 36 nm, the surface morphology and crystal quality of GaN is optimal. The in-situ reflection high energy electron diffraction (RHEED) reveals that the transition to a two-dimensional growth mode of AlN is the key to the quality of GaN. However, the thickness of AlN buffer is not so critical to the residual in-plane tensile stress in GaN grown on Si(111) by GSMBE for AlN thickness between 9 to 72 nm.

Finite element beam propagation method for analysis of plasmonic waveguide

The basic idea of the finite element beam propagation method (FE-BPM) is described. It is applied to calculate the fundamental mode of a channel plasmonic polariton (CPP) waveguide to confirm its validity. Both the field distribution and the effective index of the, fundamental mode are given by the method. The convergence speed shows the advantage and stability of this method. Then a plasmonic waveguide with a dielectric strip deposited on a metal substrate is investigated, and the group velocity is negative for the fundamental mode of this kind of waveguide. The numerical result shows that the power flow direction is reverse to that of phase velocity.

Enhancement of the far-field output power and the beam properties of plasmonic very small aperture lasers

The properties of plasmonic very small aperture lasers are shown: these integrate surface plasmon structures with very small aperture lasers. The transmission field can be confined to a spot of subwavelength width in the far field, and according to the finite difference time domain simulation results the focal length of the spot can be modulated using different ring periods. Scanning of the subwavelength gating in the far field has been realized numerically. Such a device can be used with a high-resolution far-field scanning optical microscope.