Face Recognition using Probabilistic Neural Networks

n this paper we address the problem of face detection and recognition of grey scale frontal view images. We propose a face recognition system based on probabilistic neural networks (PNN) architecture. The system is implemented using voronoi/ delaunay tessellations and template matching. Images are segmented successfully into homogeneous regions by virtue of voronoi diagram properties. Face verification is achieved using matching scores computed by correlating edge gradients of reference images. The advantage of classification using PNN models is its short training time. The correlation based template matching guarantees good classification results.

Notes on PCA, Regularization, Sparsity and Support Vector Machines

We derive a new representation for a function as a linear combination of local correlation kernels at optimal sparse locations and discuss its relation to PCA, regularization, sparsity principles and Support Vector Machines. We first review previous results for the approximation of a function from discrete data (Girosi, 1998) in the context of Vapnik"s feature space and dual representation (Vapnik, 1995). We apply them to show 1) that a standard regularization functional with a stabilizer defined in terms of the correlation function induces a regression function in the span of the feature space of classical Principal Components and 2) that there exist a dual representations of the regression function in terms of a regularization network with a kernel equal to a generalized correlation function. We then describe the main observation of the paper: the dual representation in terms of the correlation function can be sparsified using the Support Vector Machines (Vapnik, 1982) technique and this operation is equivalent to sparsify a large dictionary of basis functions adapted to the task, using a variation of Basis Pursuit De-Noising (Chen, Donoho and Saunders, 1995; see also related work by Donahue and Geiger, 1994; Olshausen and Field, 1995; Lewicki and Sejnowski, 1998). In addition to extending the close relations between regularization, Support Vector Machines and sparsity, our work also illuminates and formalizes the LFA concept of Penev and Atick (1996). We discuss the relation between our results, which are about regression, and the different problem of pattern classification.

Prognozowanie stężeń pyłku leszczyny, olszy i brzozy w powietrzu atmosferycznym na obszarze Polski

Wydział Nauk Geograficznych i Geologicznych: Instytut Geoekologii i Geoinformacji

Analysis of water-level response to rainfall and implications for recharge pathways in the Chalk aquifer, SE England

Water table response to rainfall was investigated at six sites in the Upper, Middle and Lower Chalk of southern England. Daily time series of rainfall and borehole water level were cross-corretated to investigate seasonal variations in groundwater-level response times, based on periods of 3-month duration. The time tags (in days) yielding significant correlations were compared with the average unsaturated zone thickness during each 3-month period. In general, for cases when the unsaturated zone was greater than 18 m thick, the time tag for a significant water-level response increased rapidly once the depth to the water table exceeded a critical value, which varied from site to site. For shallower water tables, a linear relationship between the depth to the water table and the water-level response time was evident. The observed variations in response time can only be partially accounted for using a diffusive model for propagation through the unsaturated matrix, suggesting that some fissure flow was occurring. The majority of rapid responses were observed during the winter/spring recharge period, when the unsaturated zone is thinnest and the unsaturated zone moisture content is highest, and were more likely to occur when the rainfall intensity exceeded 5 mm/day. At some sites, a very rapid response within 24 h of rainfall was observed in addition to the longer term responses even when the unsaturated zone was up to 64 m thick. This response was generally associated with the autumn period. The results of the cross-correlation analysis provide statistical support for the presence of fissure flow and for the contribution of multiple pathways through the unsaturated zone to groundwater recharge. (c) 2006 Elsevier B.V. All rights reserved.

Structure of AuCN determined from total neutron diffraction

The structure of gold cyanide, AuCN, has been determined at 10 and 300 K using total neutron diffraction. The structure consists of infinite -Au-(CN)-Au-(CN)-Au-(CN)- linear chains, hexagonally packed, with the gold atoms in sheets. The Au-C and Au-N bond lengths are found to be identical, with d(Au-C/N) = 1.9703(5) Angstrom at 300 K. This work supersedes a previous study, by others, which used Rietveld analysis of neutron Bragg diffraction in isolation, and found these bonds to have significantly different lengths (Deltad = 0.24 Angstrom) at 300 K. The total correlation function, T(r), at 10 and 300 K, has been modeled using information derived from total diffraction. The broadening of inter- and intrachain correlations differs markedly due to random displacements of the chains in the direction of the chain axes. This is a consequence of the relatively weak bonding between the chains. An explanation for the negative thermal expansion in the c-direction, which occurs between 10 and 300 K, is presented.

A comparison of two time-domain analysis procedures in the determination of VO2 kinetics by pseudorandom binary sequence exercise testing

The purpose of this study was to apply and compare two time-domain analysis procedures in the determination of oxygen uptake (VO2) kinetics in response to a pseudorandom binary sequence (PRBS) exercise test. PRBS exercise tests have typically been analysed in the frequency domain. However, the complex interpretation of frequency responses may have limited the application of this procedure in both sporting and clinical contexts, where a single time measurement would facilitate subject comparison. The relative potential of both a mean response time (MRT) and a peak cross-correlation time (PCCT) was investigated. This study was divided into two parts: a test-retest reliability study (part A), in which 10 healthy male subjects completed two identical PRBS exercise tests, and a comparison of the VO2 kinetics of 12 elite endurance runners (ER) and 12 elite sprinters (SR; part B). In part A, 95% limits of agreement were calculated for comparison between MRT and PCCT. The results of part A showed no significant difference between test and retest as assessed by MRT [mean (SD) 42.2 (4.2) s and 43.8 (6.9) s] or by PCCT [21.8 (3.7) s and 22.7 (4.5) s]. Measurement error (%) was lower for MRT in comparison with PCCT (16% and 25%, respectively). In part B of the study, the VO2 kinetics of ER were significantly faster than those of SR, as assessed by MRT [33.4 (3.4) s and 39.9 (7.1) s, respectively; P<0.01] and PCCT [20.9 (3.8) s and 24.8 (4.5) s; P < 0.05]. It is possible that either analysis procedure could provide a single test measurement Of VO2 kinetics; however, the greater reliability of the MRT data suggests that this method has more potential for development in the assessment Of VO2 kinetics by PRBS exercise testing.

A swarm intelligence method for feature tracking in AMV derivation

Feature tracking is a key step in the derivation of Atmospheric Motion Vectors (AMV). Most operational derivation processes use some template matching technique, such as Euclidean distance or cross-correlation, for the tracking step. As this step is very expensive computationally, often shortrange forecasts generated by Numerical Weather Prediction (NWP) systems are used to reduce the search area. Alternatives, such as optical flow methods, have been explored, with the aim of improving the number and quality of the vectors generated and the computational efficiency of the process. This paper will present the research carried out to apply Stochastic Diffusion Search, a generic search technique in the Swarm Intelligence family, to feature tracking in the context of AMV derivation. The method will be described, and we will present initial results, with Euclidean distance as reference.

Segmental dynamics in entangled linear polymer melts

We present molecular dynamics (MD) and slip-springs model simulations of the chain segmental dynamics in entangled linear polymer melts. The time-dependent behavior of the segmental orientation autocorrelation functions and mean-square segmental displacements are analyzed for both flexible and semiflexible chains, with particular attention paid to the scaling relations among these dynamic quantities. Effective combination of the two simulation methods at different coarse-graining levels allows us to explore the chain dynamics for chain lengths ranging from Z ≈ 2 to 90 entanglements. For a given chain length of Z ≈ 15, the time scales accessed span for more than 10 decades, covering all of the interesting relaxation regimes. The obtained time dependence of the monomer mean square displacements, g1(t), is in good agreement with the tube theory predictions. Results on the first- and second-order segmental orientation autocorrelation functions, C1(t) and C2(t), demonstrate a clear power law relationship of C2(t) C1(t)m with m = 3, 2, and 1 in the initial, free Rouse, and entangled (constrained Rouse) regimes, respectively. The return-to-origin hypothesis, which leads to inverse proportionality between the segmental orientation autocorrelation functions and g1(t) in the entangled regime, is convincingly verified by the simulation result of C1(t) g1(t)−1 t–1/4 in the constrained Rouse regime, where for well-entangled chains both C1(t) and g1(t) are rather insensitive to the constraint release effects. However, the second-order correlation function, C2(t), shows much stronger sensitivity to the constraint release effects and experiences a protracted crossover from the free Rouse to entangled regime. This crossover region extends for at least one decade in time longer than that of C1(t). The predicted time scaling behavior of C2(t) t–1/4 is observed in slip-springs simulations only at chain length of 90 entanglements, whereas shorter chains show higher scaling exponents. The reported simulation work can be applied to understand the observations of the NMR experiments.

Local and average structure in zinc cyanide: towards an understanding of the atomistic origin of negative thermal expansion

Neutron diffraction at 11.4 and 295 K and solid-state 67Zn NMR are used to determine both the local and average structures in the disordered, negative thermal expansion (NTE) material, Zn(CN)2. Solid-state NMR not only confirms that there is head-to-tail disorder of the C≡N groups present in the solid, but yields information about the relative abundances of the different Zn(CN)4-n(NC)n tetrahedral species, which do not follow a simple binomial distribution. The Zn(CN)4 and Zn(NC)4 species occur with much lower probabilities than are predicted by binomial theory, supporting the conclusion that they are of higher energy than the other local arrangements. The lowest energy arrangement is Zn(CN)2(NC)2. The use of total neutron diffraction at 11.4 K, with analysis of both the Bragg diffraction and the derived total correlation function, yields the first experimental determination of the individual Zn−N and Zn−C bond lengths as 1.969(2) and 2.030(2) Å, respectively. The very small difference in bond lengths, of ~0.06 Å, means that it is impossible to obtain these bond lengths using Bragg diffraction in isolation. Total neutron diffraction also provides information on both the average and local atomic displacements responsible for NTE in Zn(CN)2. The principal motions giving rise to NTE are shown to be those in which the carbon and nitrogen atoms within individual Zn−C≡N−Zn linkages are displaced to the same side of the Zn···Zn axis. Displacements of the carbon and nitrogen atoms to opposite sides of the Zn···Zn axis, suggested previously in X-ray studies as being responsible for NTE behavior, in fact make negligible contribution at temperatures up to 295 K.

A numerical model of the ionospheric signatures of time-varying magnetic reconnection: II. Measuring expansions in the ionospheric flow response

A numerical model embodying the concepts of the Cowley-Lockwood (Cowley and Lockwood, 1992, 1997) paradigm has been used to produce a simple Cowley– Lockwood type expanding flow pattern and to calculate the resulting change in ion temperature. Cross-correlation, fixed threshold analysis and threshold relative to peak are used to determine the phase speed of the change in convection pattern, in response to a change in applied reconnection. Each of these methods fails to fully recover the expansion of the onset of the convection response that is inherent in the simulations. The results of this study indicate that any expansion of the convection pattern will be best observed in time-series data using a threshold which is a fixed fraction of the peak response. We show that these methods used to determine the expansion velocity can be used to discriminate between the two main models for the convection response to a change in reconnection.

Coordinated Cluster and ground-based instrument observations of transient changes in the magnetopause boundary layer during an interval of predominantly northward IMF: relation to reconnection pulses and FTE signatures

We study a series of transient entries into the low-latitude boundary layer (LLBL) of all four Cluster spacecraft during an outbound pass through the mid-afternoon magnetopause ([X(GSM), Y(GSM), Z(GSM)] approximate to [2, 7, 9] R(E)). The events take place during an interval of northward IMF, as seen in the data from the ACE satellite and lagged by a propagation delay of 75 min that is well-defined by two separate studies: (1) the magnetospheric variations prior to the northward turning (Lockwood et al., 2001, this issue) and (2) the field clock angle seen by Cluster after it had emerged into the magnetosheath (Opgenoorth et al., 2001, this issue). With an additional lag of 16.5 min, the transient LLBL events cor-relate well with swings of the IMF clock angle (in GSM) to near 90degrees. Most of this additional lag is explained by ground-based observations, which reveal signatures of transient reconnection in the pre-noon sector that then take 10-15 min to propagate eastward to 15 MLT, where they are observed by Cluster. The eastward phase speed of these signatures agrees very well with the motion deduced by the cross-correlation of the signatures seen on the four Cluster spacecraft. The evidence that these events are reconnection pulses includes: transient erosion of the noon 630 nm (cusp/cleft) aurora to lower latitudes; transient and travelling enhancements of the flow into the polar cap, imaged by the AMIE technique; and poleward-moving events moving into the polar cap, seen by the EISCAT Svalbard Radar (ESR). A pass of the DMSP-F15 satellite reveals that the open field lines near noon have been opened for some time: the more recently opened field lines were found closer to dusk where the flow transient and the poleward-moving event intersected the satellite pass. The events at Cluster have ion and electron characteristics predicted and observed by Lockwood and Hapgood (1998) for a Flux Transfer Event (FTE), with allowance for magnetospheric ion reflection at Alfvenic disturbances in the magnetopause reconnection layer. Like FTEs, the events are about 1 R(E) in their direction of motion and show a rise in the magnetic field strength, but unlike FTEs, in general, they show no pressure excess in their core and hence, no characteristic bipolar signature in the boundary-normal component. However, most of the events were observed when the magnetic field was southward, i.e. on the edge of the interior magnetic cusp, or when the field was parallel to the magnetic equatorial plane. Only when the satellite begins to emerge from the exterior boundary (when the field was northward), do the events start to show a pressure excess in their core and the consequent bipolar signature. We identify the events as the first observations of FTEs at middle altitudes.

Dayside ionospheric convection changes in response to long-period interplanetary Magnetic field oscillations: Determination of the ionospheric phase velocity

Ground magnetic field perturbations recorded by the CANOPUS magnetometer network in the 7 to 13 MLT sector are used to examine how reconfigurations of the dayside polar ionospheric flow take place in response to north-south changes of the IMF. During the 6-hour interval in question IMF Bz oscillates between ±7 nT with about a 1-hour period. Corresponding variations in the ground magnetic disturbance are observed which we infer are due to changes in ionospheric flow. Cross correlation of the data obtained from two ground stations at 73.5° magnetic latitude, but separated by ∼2 hours in MLT, shows that changes in the flow are initiated in the prenoon sector (∼10 MLT) and then spread outward toward dawn and dusk with a phase speed of ∼5 km s−1 over the longitude range ∼8 to 12 MLT, slowing to ∼2 km s−1 outside this range. Cross correlating the data from these ground stations with IMP 8 IMF Bz records produces a MLT variation in the ground response delay relative to the IMF which is compatible with these deduced phase speeds. We interpret these observations in terms of the ionospheric response to the onset, expansion and decay of magnetic reconnection at the dayside magnetopause.

Response time of the high-latitude dayside ionosphere to sudden changes in the north-south component of the IMF

The time scale of the response of the high-latitude dayside ionospheric flow to changes in the North-South component of the interplanetary magnetic field (IMF) has been investigated by examining the time delays between corresponding sudden changes. Approximately 40 h of simultaneous IMF and ionospheric flow data have been examined, obtained by the AMPTE-UKS and -IRM spacecraft and the EISCAT “Polar” experiment, respectively, in which 20 corresponding sudden changes have been identified. Ten of these changes were associated with southward turnings of the IMF, and 10 with northward turnings. It has been found that the corresponding flow changes occurred simultaneously over the whole of the “Polar” field-of-view, extending more than 2° in invariant latitude, and that the ionospheric response delay following northward turnings is the same as that following southward turnings, though the form of the response is different in the two cases. The shortest response time, 5.5 ± 3.2 min, is found in the early- to mid-afternoon sector, increasing to 9.5 ± 3.0 min in the mid-morning sector, and to 9.5 ± 3.1 min near to dusk. These times represent the delays in the appearance of perturbed flows in the “Polar” field-of-view following the arrival of IMF changes at the subsolar magnetopause. Overall, the results agree very well with those derived by Etemadi et al. (1988, Planet. Space Sci.36, 471) from a general cross-correlation analysis of the IMF Bz and “Polar” beam-swinging vector flow data.

The chaotic frequency hopping signals for MIMO radar

In multiple-input multiple-output (MIMO) radar systems, the transmitters emit orthogonal waveforms to increase the spatial resolution. New frequency hopping (FH) codes based on chaotic sequences are proposed. The chaotic sequences have the characteristics of good encryption, anti-jamming properties and anti-intercept capabilities. The main idea of chaotic FH is based on queuing theory. According to the sensitivity to initial condition, these sequences can achieve good Hamming auto-correlation while also preserving good average correlation. Simulation results show that the proposed FH signals can achieve lower autocorrelation side lobe level and peak cross-correlation level with the increasing of iterations. Compared to the LFM signals, this sequence has higher range-doppler resolution.

Video stimuli reduce object-directed imitation accuracy: a novel two-person motion-tracking approach

Imitation is an important form of social behavior, and research has aimed to discover and explain the neural and kinematic aspects of imitation. However, much of this research has featured single participants imitating in response to pre-recorded video stimuli. This is in spite of findings that show reduced neural activation to video vs. real life movement stimuli, particularly in the motor cortex. We investigated the degree to which video stimuli may affect the imitation process using a novel motion tracking paradigm with high spatial and temporal resolution. We recorded 14 positions on the hands, arms, and heads of two individuals in an imitation experiment. One individual freely moved within given parameters (moving balls across a series of pegs) and a second participant imitated. This task was performed with either simple (one ball) or complex (three balls) movement difficulty, and either face-to-face or via a live video projection. After an exploratory analysis, three dependent variables were chosen for examination: 3D grip position, joint angles in the arm, and grip aperture. A cross-correlation and multivariate analysis revealed that object-directed imitation task accuracy (as represented by grip position) was reduced in video compared to face-to-face feedback, and in complex compared to simple difficulty. This was most prevalent in the left-right and forward-back motions, relevant to the imitator sitting face-to-face with the actor or with a live projected video of the same actor. The results suggest that for tasks which require object-directed imitation, video stimuli may not be an ecologically valid way to present task materials. However, no similar effects were found in the joint angle and grip aperture variables, suggesting that there are limits to the influence of video stimuli on imitation. The implications of these results are discussed with regards to previous findings, and with suggestions for future experimentation.