Incorporating curvature information into on-line learning

We analyse the dynamics of a number of second order on-line learning algorithms training multi-layer neural networks, using the methods of statistical mechanics. We first consider on-line Newton's method, which is known to provide optimal asymptotic performance. We determine the asymptotic generalization error decay for a soft committee machine, which is shown to compare favourably with the result for standard gradient descent. Matrix momentum provides a practical approximation to this method by allowing an efficient inversion of the Hessian. We consider an idealized matrix momentum algorithm which requires access to the Hessian and find close correspondence with the dynamics of on-line Newton's method. In practice, the Hessian will not be known on-line and we therefore consider matrix momentum using a single example approximation to the Hessian. In this case good asymptotic performance may still be achieved, but the algorithm is now sensitive to parameter choice because of noise in the Hessian estimate. On-line Newton's method is not appropriate during the transient learning phase, since a suboptimal unstable fixed point of the gradient descent dynamics becomes stable for this algorithm. A principled alternative is to use Amari's natural gradient learning algorithm and we show how this method provides a significant reduction in learning time when compared to gradient descent, while retaining the asymptotic performance of on-line Newton's method.

Compositional semantics of dataflow networks with query-driven communication of exact values

We develop and study the concept of dataflow process networks as used for exampleby Kahn to suit exact computation over data types related to real numbers, such as continuous functions and geometrical solids. Furthermore, we consider communicating these exact objectsamong processes using protocols of a query-answer nature as introduced in our earlier work. This enables processes to provide valid approximations with certain accuracy and focusing on certainlocality as demanded by the receiving processes through queries. We define domain-theoretical denotational semantics of our networks in two ways: (1) directly, i. e. by viewing the whole network as a composite process and applying the process semantics introduced in our earlier work; and (2) compositionally, i. e. by a fixed-point construction similarto that used by Kahn from the denotational semantics of individual processes in the network. The direct semantics closely corresponds to the operational semantics of the network (i. e. it iscorrect) but very difficult to study for concrete networks. The compositional semantics enablescompositional analysis of concrete networks, assuming it is correct. We prove that the compositional semantics is a safe approximation of the direct semantics. Wealso provide a method that can be used in many cases to establish that the two semantics fully coincide, i. e. safety is not achieved through inactivity or meaningless answers. The results are extended to cover recursively-defined infinite networks as well as nested finitenetworks. A robust prototype implementation of our model is available.

Fast reconstruction of harmonic functions from Cauchy data using integral equation techniques

We consider the problem of stable determination of a harmonic function from knowledge of the solution and its normal derivative on a part of the boundary of the (bounded) solution domain. The alternating method is a procedure to generate an approximation to the harmonic function from such Cauchy data and we investigate a numerical implementation of this procedure based on Fredholm integral equations and Nyström discretization schemes, which makes it possible to perform a large number of iterations (millions) with minor computational cost (seconds) and high accuracy. Moreover, the original problem is rewritten as a fixed point equation on the boundary, and various other direct regularization techniques are discussed to solve that equation. We also discuss how knowledge of the smoothness of the data can be used to further improve the accuracy. Numerical examples are presented showing that accurate approximations of both the solution and its normal derivative can be obtained with much less computational time than in previous works.

Control of pattern formation by time-delay feedback with global and local contributions

We consider the suppression of spatiotemporal chaos in the complex GinzburgLandau equation by a combined global and local time-delay feedback. Feedback terms are implemented as a control scheme, i.e., they are proportional to the difference between the time-delayed state of the system and its current state. We perform a linear stability analysis of uniform oscillations with respect to space-dependent perturbations and compare with numerical simulations. Similarly, for the fixed-point solution that corresponds to amplitude death in the spatially extended system, a linear stability analysis with respect to space-dependent perturbations is performed and complemented by numerical simulations. © 2010 Elsevier B.V. All rights reserved.

Refractive index sensitivity characteristics near the dispersion turning point of the multimode microfiber-based Mach–Zehnder interferometer

The turning point of the refractive index (RI) sensitivity based on the multimode microfiber (MMMF) in-line Mach–Zehnder interferometer (MZI) is observed. By tracking the resonant wavelength shift of the MZI generated between the HE11 and HE12 modes in the MMMF, the surrounding RI (SRI) could be detected. Theoretical analysis demonstrates that the RI sensitivity will reach ±∞ on either side of the turning point due to the group effective RI difference (퐺) approaching zero. Significantly, the positive sensitivity exists in a very wide fiber diameter range, while the negative sensitivity can be achieved in a narrow diameter range of only 0.3 μm. Meanwhile, the experimental sensitivities and variation trend at different diameters exhibit high consistency with the theoretical results. High RI sensitivity of 10777.8 nm/RIU (RI unit) at the fiber diameter of 4.6 μm and the RI around 1.3334 is realized. The discovery of the sensitivity turning points has great significance on trace detection due to the possibility of ultrahigh RI sensitivity.

Spatially clustered associations in health GIS

Overlaying maps using a desktop GIS is often the first step of a multivariate spatial analysis. The potential of this operation has increased considerably as data sources and Web services to manipulate them are becoming widely available via the Internet. Standards from the OGC enable such geospatial mashups to be seamless and user driven, involving discovery of thematic data. The user is naturally inclined to look for spatial clusters and correlation of outcomes. Using classical cluster detection scan methods to identify multivariate associations can be problematic in this context, because of a lack of control on or knowledge about background populations. For public health and epidemiological mapping, this limiting factor can be critical but often the focus is on spatial identification of risk factors associated with health or clinical status. Spatial entropy index HSu for the ScankOO analysis of the hypothetical dataset using a vicinity which is fixed by the number of points without distinction between their labels. (The size of the labels is proportional to the inverse of the index) In this article we point out that this association itself can ensure some control on underlying populations, and develop an exploratory scan statistic framework for multivariate associations. Inference using statistical map methodologies can be used to test the clustered associations. The approach is illustrated with a hypothetical data example and an epidemiological study on community MRSA. Scenarios of potential use for online mashups are introduced but full implementation is left for further research.

Point-by-point inscription of first-order fiber Bragg grating for C-band applications

The influence of the fiber geometry on the point-by-point inscription of fiber Bragg gratings using a femtosecond laser is highlighted. Fiber Bragg gratings with high spectral quality and strong first-order Bragg resonances within the C-band are achieved by optimizing the inscription process. Large birefringence (1.2×10-4) and high degree of polarizationdependent index modulation are observed in these gratings. Potential applications of these gratings in resonators are further illustrated.

Dual-peak long-period fiber gratings with enhanced refractive index sensitivity by finely tailored mode dispersion that uses the light cladding etching technique

We have experimentally investigated the mode dispersion property and refractive index sensitivity of dual-peak long-period fiber gratings (LPGs) that were sensitized by hydrofluoric acid (HF) etching. The nature of the coupled cladding modes close to the dispersion turning point makes the dual-peak LPGs ultrasensitive to cladding property, permitting a fine tailoring of the mode dispersion and index sensitivity by the light cladding etching method using HF acid of only 1% concentration. As an implementation of an optical biosensor, the etched device was used to detect the concentration of hemoglobin protein in a sugar solution, showing a sensitivity as high as 20 nm/1%.

Point-by-point inscription of first-order fiber Bragg grating for C-band applications

The influence of the fiber geometry on the point-by-point inscription of fiber Bragg gratings using a femtosecond laser is highlighted. Fiber Bragg gratings with high spectral quality and strong first-order Bragg resonances within the C-band are achieved by optimizing the inscription process. Large birefringence (1.2x10-4) and high degree of polarizationdependent index modulation are observed in these gratings. Potential applications of these gratings in resonators are further illustrated. © 2007 Optical Society of America.

Fiber laser incorporating an intracavity micro-channel for refractive index and temperature sensing

In this letter, we present a standard linear cavity fiber laser incorporating a microchannel for refractive index (RI) and temperature sensing. The microchannel of ~6µm width created by femtosecond laser aided chemical etching provides an access to the external liquid; therefore, the laser cavity loss changes with the liquids of different RIs. Thus, at a fixed pump power, the output laser power will vary with the change of RI in the microchannel. The results show that the proposed sensing system has a linear response to both the surrounding medium RI and temperature. The RI sensitivity of the laser system is on the order of 10-3, while the temperature sensitivity is about 0.02 C. Both sensitivities could be further enhanced by employing a more sensitive photodetector and using higher pump power.

On bandwidth request mechanism with piggyback in fixed IEEE 802.16 networks

This paper investigates the random channel access mechanism specified in the IEEE 802.16 standard for the uplink traffic in a Point-to-MultiPoint (PMP) network architecture. An analytical model is proposed to study the impacts of the channel access parameters, bandwidth configuration and piggyback policy on the performance. The impacts of physical burst profile and non-saturated network traffic are also taken into account in the model. Simulations validate the proposed analytical model. It is observed that the bandwidth utilization can be improved if the bandwidth for random channel access can be properly configured according to the channel access parameters, piggyback policy and network traffic.

Performance comparison of OFDM bandwidth request schemes in fixed IEEE 802.16 networks

IEEE 802.16 standard specifies two contention based bandwidth request schemes working with OFDM physical layer specification in point-to-multipoint (PMP) architecture, the mandatory one used in region-full and the optional one used in region-focused. This letter presents a unified analytical model to study the bandwidth efficiency and channel access delay performance of the two schemes. The impacts of access parameters, available bandwidth and subchannelization have been taken into account. The model is validated by simulations. The mandatory scheme is observed to perform closely to the optional one when subchannelization is active for both schemes.

Temperature-referenced high-sensitivity point-probe optical fiber chem-sensors based on cladding etched fiber Bragg gratings

Point-probe optical fiber chem-sensors have been implemented using cladding etched fiber Bragg gratings. The sensors possess refractive index sensing capability that can be utilized to measure chemical concentrations. The Bragg wavelength shift reaches 8 nm when the index of surrounding medium changes from 1.33 to 1.44, giving maximum sensitivity more than 10 times higher than that of previously reported devices. More importantly, the dual-grating configuration of the point-probe sensors offers a temperature reference function, permitting accurate measurement of refractive index encoded chemical concentrations.

Temperature-referenced high-sensitivity point-probe optical fiber chem-sensors based on cladding etched fiber Bragg gratings

Point-probe optical fiber chem-sensors have been implemented using cladding etched fiber Bragg gratings. The sensors possess refractive index sensing capability that can be utilized to measure chemical concentrations. The Bragg wavelength shift reaches 8 nm when the index of surrounding medium changes from 1.33 to 1.44, giving maximum sensitivity more than 10 times higher than that of previously reported devices. More importantly, the dual-grating configuration of the point-probe sensors offers a temperature reference function, permitting accurate measurement of refractive index encoded chemical concentrations.

Dual-peak long-period fiber gratings with enhanced refractive index sensitivity by finely tailored mode dispersion that uses the light cladding etching technique

We have experimentally investigated the mode dispersion property and refractive index sensitivity of dual-peak long-period fiber gratings (LPGs) that were sensitized by hydrofluoric acid (HF) etching. The nature of the coupled cladding modes close to the dispersion turning point makes the dual-peak LPGs ultrasensitive to cladding property, permitting a fine tailoring of the mode dispersion and index sensitivity by the light cladding etching method using HF acid of only 1% concentration. As an implementation of an optical biosensor, the etched device was used to detect the concentration of hemoglobin protein in a sugar solution, showing a sensitivity as high as 20 nm/1%. © 2007 Optical Society of America.