Neural Structures Using the Eigenstates of a Quantum Harmonic Oscillator

The main result of the paper is the use of orthogonal Hermite polynomials as the basis functions of feedforward neural networks. The proposed neural networks have some interesting properties: (i) the basis functions are invariant under the Fourier transform, subject only to a change of scale, (ii) the basis functions are the eigenstates of the quantum harmonic oscillator, and stem from the solution of Schrödinger's diffusion equation. The proposed feed-forward neural networks demonstrate the particle-wave nature of information and can be used in nonparametric estimation. Possible applications of the proposed neural networks include function approximation, image processing and system modelling.     

Hand dexterity assessment based on mouse pointer trajectory measurements in children with learning disabilities

Universal Access in the Information Society - We present an investigation of the systematic differences in upper limb motor skills between children with learning disabilities (LDs) and children...     

Colloidal Quantum Dot Infrared Lasers Featuring Sub-Single-Exciton Threshold and Very High Gain

The use of colloidal quantum dots (CQDs) as a gain medium in infrared laser devices has been underpinned by the need for high pumping intensities, very short gain lifetimes, and low gain coefficients. Here, PbS/PbSSe core/alloyed-shell CQDs are employed as an infrared gain medium that results in highly suppressed Auger recombination with a lifetime of 485 ps, lowering the amplified spontaneous emission (ASE) threshold down to 300 µJ cm⁻², and showing a record high net modal gain coefficient of 2180 cm⁻¹. By doping these engineered core/shell CQDs up to nearly filling the first excited state, a significant reduction of optical gain threshold is demonstrated, measured by transient absorption, to an average-exciton population-per-dot 〈N^th〉_g of 0.45 due to bleaching of the ground state absorption. This in turn have led to a fivefold reduction in ASE threshold at 〈N^th〉_ASE = 0.70 excitons-per-dot, associated with a gain lifetime of 280 ps. Finally, these heterostructured QDs are used to achieve near-infrared lasing at 1670 nm at a pump fluences corresponding to sub-single-exciton-per-dot threshold (〈N^th〉_Las = 0.87). This work brings infrared CQD lasing thresholds on par to their visible counterparts, and paves the way toward solution-processed infrared laser diodes.     

Extended Kalman and Particle Filtering for sensor fusion in motion control of mobile robots

Motion control of mobile robots and efficient trajectory tracking is usually based on prior estimation of the robots’ state vector. To this end Gaussian and nonparametric filters (state estimators from position measurements) have been developed. In this paper the Extended Kalman Filter which assumes Gaussian measurement noise is compared to the Particle Filter which does not make any assumption on the measurement noise distribution. As a case study the estimation of the state vector of a mobile robot is used, when measurements are available from both odometric and sonar sensors. It is shown that in this kind of sensor fusion problem the Particle Filter has better performance than the Extended Kalman Filter, at the cost of more demanding computations.     

Stochastic Processes and Neuronal Modelling: Quantum Harmonic Oscillator Dynamics in Neural Structures

This paper studies neural structures with weights that follow the model of the quantum harmonic oscillator (Q.H.O.). The proposed neural networks have stochastic weights which are calculated from the solution of Schrödinger’s equation under the assumption of a parabolic (harmonic) potential. These weights correspond to diffusing particles, which interact to each other as the theory of Brownian motion (Wiener process) predicts. The learning of the stochastic weights (convergence of the diffusing particles to an equilibrium) is analyzed. In the case of associative memories the proposed neural model results in an exponential increase of patterns storage capacity (number of attractors). Finally, it is shown that conventional neural networks and learning algorithms based on error gradient can be conceived as a subset of the proposed quantum neural structures. Thus, the complementarity between classical and quantum physics is also validated in the field of neural computation.     

A Study in Efficiency and Modality Usage in Multimodal Form Filling Systems

The usage patterns of speech and visual input modes are investigated as a function of relative input mode efficiency for both desktop and personal digital assistant (PDA) working environments. For this purpose the form-filling part of a multimodal dialogue system is implemented and evaluated; three multimodal modes of interaction are implemented: ldquoClick-to-Talk,rdquo ldquoOpen-Mike,rdquo and ldquoModality-Selection.rdquo ldquoModality-Selectionrdquo implements an adaptive interface where the system selects the most efficient input mode at each turn, effectively alternating between a ldquoClick-to-Talkrdquo and ldquoOpen-Mikerdquo interaction style as proposed in ldquoModality tracking in the multimodal Bell Labs Communicator,rdquo in Proceedings of the Automatic Speech Recognition and Understanding Workshop, by A. Potamianos, , 2003. The multimodal systems are evaluated and compared with the unimodal systems. Objective and subjective measures used include task completion, task duration, turn duration, and overall user satisfaction. Turn duration is broken down into interaction time and inactivity time to better measure the efficiency of each input mode. Duration statistics and empirical probability density functions are computed as a function of interaction context and user. Results show that the multimodal systems outperform the unimodal systems in terms of objective and subjective criteria. Also, users tend to use the most efficient input mode at each turn; however, biases towards the default input modality and a general bias towards the speech modality also exists. Results demonstrate that although users exploit some of the available synergies in multimodal dialogue interaction, further efficiency gains can be achieved by designing adaptive interfaces that fully exploit these synergies.     

Modeling of fermentative hydrogen production from sweet sorghum extract based on modified ADM1

The Anaerobic digestion model 1 (ADM1) framework can be used to predict fermentative hydrogen production, since the latter is directly related to the acidogenic stage of the anaerobic digestion process. In this study, the ADM1 model framework was used to simulate and predict the process of fermentative hydrogen production from the extractable sugars of sweet sorghum biomass. Kinetic parameters for sugars’ consumption and yield coefficients of acetic, propionic and butyric acid production were estimated using the experimental data obtained from the steady states of a CSTR. Batch experiments were used for kinetic parameter validation. Since the ADM1 does not account for metabolic products such as lactic acid and ethanol that are crucial during the fermentative hydrogen production process, the structure of the model was modified to include lactate and ethanol among the metabolites and to improve the predictions. The modified ADM1 simulated satisfactorily batch experiments although further modifications could be made in order to further improve the predictions for the hydrogenogenic process.     

Joint face and head tracking inside multi-camera smart rooms

The paper introduces a novel detection and tracking system that provides both frame-view and world-coordinate human location information, based on video from multiple synchronized and calibrated cameras with overlapping fields of view. The system is developed and evaluated for the specific scenario of a seminar lecturer presenting in front of an audience inside a “smart room”, its aim being to track the lecturer’s head centroid in the three-dimensional (3D) space and also yield two-dimensional (2D) face information in the available camera views. The proposed approach is primarily based on a statistical appearance model of human faces by means of well-known AdaBoost-like face detectors, extended to address the head pose variation observed in the smart room scenario of interest. The appearance module is complemented by two novel components and assisted by a simple tracking drift detection mechanism. The first component of interest is the initialization module, which employs a spatio-temporal dynamic programming approach with appropriate penalty functions to obtain optimal 3D location hypotheses. The second is an adaptive subspace learning based 2D tracking scheme with a novel forgetting mechanism, introduced to reduce tracking drift and increase robustness. System performance is benchmarked on an extensive database of realistic human interaction in the lecture smart room scenario, collected as part of the European integrated project “CHIL”. The system consistently achieves excellent tracking precision, with a 3D mean tracking error of less than 16 cm, and is demonstrated to outperform four alternative tracking schemes. Furthermore, the proposed system performs relatively well in detecting frontal and near-frontal faces in the available frame views. This work was performed while Zhenqiu Zhang was on a summer internship with the Human Language Technology Department at the IBM T.J. Watson Research Center.