Towards a parallel computationally efficient approach to scaling up data stream classification

Advances in hardware technologies allow to capture and process data in real-time and the resulting high throughput data streams require novel data mining approaches. The research area of Data Stream Mining (DSM) is developing data mining algorithms that allow us to analyse these continuous streams of data in real-time. The creation and real-time adaption of classification models from data streams is one of the most challenging DSM tasks. Current classifiers for streaming data address this problem by using incremental learning algorithms. However, even so these algorithms are fast, they are challenged by high velocity data streams, where data instances are incoming at a fast rate. This is problematic if the applications desire that there is no or only a very little delay between changes in the patterns of the stream and absorption of these patterns by the classifier. Problems of scalability to Big Data of traditional data mining algorithms for static (non streaming) datasets have been addressed through the development of parallel classifiers. However, there is very little work on the parallelisation of data stream classification techniques. In this paper we investigate K-Nearest Neighbours (KNN) as the basis for a real-time adaptive and parallel methodology for scalable data stream classification tasks.

Chaos in high performance digital robot controllers

Active robot force control requires some form of dynamic inner loop control for stability. The author considers the implementation of position-based inner loop control on an industrial robot fitted with encoders only. It is shown that high gain velocity feedback for such a robot, which is effectively stationary when in contact with a stiff environment, involves problems beyond the usual caveats on the effects of unknown environment stiffness. It is shown that it is possible for the controlled joint to become chaotic at very low velocities if encoder edge timing data are used for velocity measurement. The results obtained indicate that there is a lower limit on controlled velocity when encoders are the only means of joint measurement. This lower limit to speed is determined by the desired amount of loop gain, which is itself determined by the severity of the nonlinearities present in the drive system.

The effects of explicit versus parameterized convection on the MJO in a large-domain high-resolution tropical case study. Part I: Characterization of large-scale organization and propagation

High-resolution simulations over a large tropical domain (∼20◦S–20◦N and 42◦E–180◦E) using both explicit and parameterized convection are analyzed and compared to observations during a 10-day case study of an active Madden-Julian Oscillation (MJO) event. The parameterized convection model simulations at both 40 km and 12 km grid spacing have a very weak MJO signal and little eastward propagation. A 4 km explicit convection simulation using Smagorinsky subgrid mixing in the vertical and horizontal dimensions exhibits the best MJO strength and propagation speed. 12 km explicit convection simulations also perform much better than the 12 km parameterized convection run, suggesting that the convection scheme, rather than horizontal resolution, is key for these MJO simulations. Interestingly, a 4 km explicit convection simulation using the conventional boundary layer scheme for vertical subgrid mixing (but still using Smagorinsky horizontal mixing) completely loses the large-scale MJO organization, showing that relatively high resolution with explicit convection does not guarantee a good MJO simulation. Models with a good MJO representation have a more realistic relationship between lower-free-tropospheric moisture and precipitation, supporting the idea that moisture-convection feedback is a key process for MJO propagation. There is also increased generation of available potential energy and conversion of that energy into kinetic energy in models with a more realistic MJO, which is related to larger zonal variance in convective heating and vertical velocity, larger zonal temperature variance around 200 hPa, and larger correlations between temperature and ascent (and between temperature and diabatic heating) between 500–400 hPa.

High sensitivity recording of afferent nerve activity using ultra-compliant microchannel electrodes: an acutein vivovalidation

Neuroprostheses interfaced with transected peripheral nerves are technological routes to control robotic limbs as well as convey sensory feedback to patients suffering from traumatic neural injuries or degenerative diseases. To maximize the wealth of data obtained in recordings, interfacing devices are required to have intrafascicular resolution and provide high signal-to-noise ratio (SNR) recordings. In this paper, we focus on a possible building block of a three-dimensional regenerative implant: a polydimethylsiloxane (PDMS) microchannel electrode capable of highly sensitive recordings in vivo. The PDMS 'micro-cuff' consists of a 3.5 mm long (100 µm × 70 µm cross section) microfluidic channel equipped with five evaporated Ti/Au/Ti electrodes of sub-100 nm thickness. Individual electrodes have average impedance of 640 ± 30 kΩ with a phase angle of −58 ± 1 degrees at 1 kHz and survive demanding mechanical handling such as twisting and bending. In proof-of-principle acute implantation experiments in rats, surgically teased afferent nerve strands from the L5 dorsal root were threaded through the microchannel. Tactile stimulation of the skin was reliably monitored with the three inner electrodes in the device, simultaneously recording signal amplitudes of up to 50 µV under saline immersion. The overall SNR was approximately 4. A small but consistent time lag between the signals arriving at the three electrodes was observed and yields a fibre conduction velocity of 30 m s−1. The fidelity of the recordings was verified by placing the same nerve strand in oil and recording activity with hook electrodes. Our results show that PDMS microchannel electrodes open a promising technological path to 3D regenerative interfaces.

On the generalized eigenvalue problem for the Rossby wave vertical velocity in the presence of mean flow and topography

In a series of papers, Killworth and Blundell have proposed to study the effects of a background mean flow and topography on Rossby wave propagation by means of a generalized eigenvalue problem formulated in terms of the vertical velocity, obtained from a linearization of the primitive equations of motion. However, it has been known for a number of years that this eigenvalue problem contains an error, which Killworth was prevented from correcting himself by his unfortunate passing and whose correction is therefore taken up in this note. Here, the author shows in the context of quasigeostrophic (QG) theory that the error can ulti- mately be traced to the fact that the eigenvalue problem for the vertical velocity is fundamentally a non- linear one (the eigenvalue appears both in the numerator and denominator), unlike that for the pressure. The reason that this nonlinear term is lacking in the Killworth and Blundell theory comes from neglecting the depth dependence of a depth-dependent term. This nonlinear term is shown on idealized examples to alter significantly the Rossby wave dispersion relation in the high-wavenumber regime but is otherwise irrelevant in the long-wave limit, in which case the eigenvalue problems for the vertical velocity and pressure are both linear. In the general dispersive case, however, one should first solve the generalized eigenvalue problem for the pressure vertical structure and, if needed, diagnose the vertical velocity vertical structure from the latter.

The vertical structure of oceanic Rossby waves: a comparison of high-resolution model data to theoretical vertical structures

Tests of the new Rossby wave theories that have been developed over the past decade to account for discrepancies between theoretical wave speeds and those observed by satellite altimeters have focused primarily on the surface signature of such waves. It appears, however, that the surface signature of the waves acts only as a rather weak constraint, and that information on the vertical structure of the waves is required to better discriminate between competing theories. Due to the lack of 3-D observations, this paper uses high-resolution model data to construct realistic vertical structures of Rossby waves and compares these to structures predicted by theory. The meridional velocity of a section at 24° S in the Atlantic Ocean is pre-processed using the Radon transform to select the dominant westward signal. Normalized profiles are then constructed using three complementary methods based respectively on: (1) averaging vertical profiles of velocity, (2) diagnosing the amplitude of the Radon transform of the westward propagating signal at different depths, and (3) EOF analysis. These profiles are compared to profiles calculated using four different Rossby wave theories: standard linear theory (SLT), SLT plus mean flow, SLT plus topographic effects, and theory including mean flow and topographic effects. Our results support the classical theoretical assumption that westward propagating signals have a well-defined vertical modal structure associated with a phase speed independent of depth, in contrast with the conclusions of a recent study using the same model but for different locations in the North Atlantic. The model structures are in general surface intensified, with a sign reversal at depth in some regions, notably occurring at shallower depths in the East Atlantic. SLT provides a good fit to the model structures in the top 300 m, but grossly overestimates the sign reversal at depth. The addition of mean flow slightly improves the latter issue, but is too surface intensified. SLT plus topography rectifies the overestimation of the sign reversal, but overestimates the amplitude of the structure for much of the layer above the sign reversal. Combining the effects of mean flow and topography provided the best fit for the mean model profiles, although small errors at the surface and mid-depths are carried over from the individual effects of mean flow and topography respectively. Across the section the best fitting theory varies between SLT plus topography and topography with mean flow, with, in general, SLT plus topography performing better in the east where the sign reversal is less pronounced. None of the theories could accurately reproduce the deeper sign reversals in the west. All theories performed badly at the boundaries. The generalization of this method to other latitudes, oceans, models and baroclinic modes would provide greater insight into the variability in the ocean, while better observational data would allow verification of the model findings.

EISCAT observations of ion composition and temperature anisotropy in the high-latitude F-region

The papers by Winser et al. [(1990) J. atmos. terr. Phys.52, 501] and Häggström and Collis [(1990) J. atmos. terr. Phys.52, 519] used plasma flows and ion temperatures, as measured by the EISCAT tristatic incoherent scatter radar, to investigate changes in the ion composition of the ionospheric F-layer at high latitudes, in response to increases in the speed of plasma convection. These studies reported that the ion composition rapidly changed from mainly O+ to almost completely (>90%) molecular ions, following rapid increases in ion drift speed by >1 km s−1. These changes appeared inconsisent with theoretical considerations of the ion chemistry, which could not account for the large fractions of molecular ions inferred from the obsevations. In this paper, we discuss two causes of this discrepancy. First, we reevaluate the theoretical calculations for chemical equilibrium and show that, if we correct the derived temperatures for the effect of the molecular ions, and if we employ more realistic dependences of the reaction rates on the ion temperature, the composition changes derived for the faster convection speeds can be explained. For the Winser et al. observations with the radar beam at an aspect angle of ϕ = 54.7° to the geomagnetic field, we now compute a change to 89% molecular ions in < 2 min, in response to the 3 km s−1 drift. This is broadly consistent with the observations. But for the two cases considered by Häggström and Collis, looking along the field line (ϕ = 0°), we compute the proportion of molecular ions to be only 4 and 16% for the observed plasma drifts of 1.2 and 1.6 km s−1, respectively. These computed proportions are much smaller than those derived experimentally (70 and 90%). We attribute the differences to the effects of non-Maxwellian, anisotropic ion velocity distribution functions. We also discuss the effect of ion composition changes on the various radar observations that report anisotropies of ion temperature.

Measuring ion temperatures and studying the ion energy balance in the high-latitude ionosphere

Data are presented for a nighttime ion heating event observed by the EISCAT radar on 16 December 1988. In the experiment, the aspect angle between the radar beam and the geomagnetic field was fixed at 54.7°, which avoids any ambiguity in derived ion temperature caused by anisotropy in the ion velocity distribution function. The data were analyzed with an algorithm which takes account of the non-Maxwellian line-of-sight ion velocity distribution. During the heating event, the derived spectral distortion parameter (D∗) indicated that the distribution function was highly distorted from a Maxwellian form when the ion drift increased to 4 km s−1. The true three-dimensional ion temperature was used in the simplified ion balance equation to compute the ion mass during the heating event. The ion composition was found to change from predominantly O4 to mainly molecular ions. A theoretical analysis of the ion composition, using the MSIS86 model and published values of the chemical rate coefficients, accounts for the order-of-magnitude increase in the atomic/molecular ion ratio during the event, but does not successfully explain the very high proportion of molecular ions that was observed.

Characteristics of the high-latitude trough

The EISCAT radar has provided data for a comprehensive study of the high-latitude trough in electron concentration, which occurs in the auroral zone. In this paper the characteristics of the trough are illustrated, the method of its formation is outlined and important features of the trough are described. A large upward velocity along the geomagnetic field line is shown to play a significant role in the formation of the trough. The large ion-neutral difference velocities which initiate the formation of the trough may also drive the plasma into a non-thermal state which should be taken into account during the analysis of incoherent scatter data.

Comparisons between EISCAT observations and model calculations of the high latitude ionosphere

Calculations using a numerical model of the convection dominated high latitude ionosphere are compared with observations made by EISCAT as part of the UK-POLAR Special Programme. The data used were for 24–25 October 1984, which was characterized by an unusually steady IMF, with Bz < 0 and By > 0; in the calculations it was assumed that a steady IMF implies steady convection conditions. Using the electric field models of Heppner and Maynard (1983) appropriate to By > 0 and precipitation data taken from Spiroet al. (1982), we calculated the velocities and electron densities appropriate to the EISCAT observations. Many of the general features of the velocity data were reproduced by the model. In particular, the phasing of the change from eastward to westward flow in the vicinity of the Harang discontinuity, flows near the dayside throat and a region of slow flow at higher latitudes near dusk were well reproduced. In the afternoon sector modelled velocity values were significantly less than those observed. Electron density calculations showed good agreement with EISCAT observations near the F-peak, but compared poorly with observations near 211 km. In both cases, the greatest disagreement occurred in the early part of the observations, where the convection pattern was poorly known and showed some evidence of long term temporal change. Possible causes for the disagreement between observations and calculations are discussed and shown to raise interesting and, as yet, unresolved questions concerning the interpretation of the data. For the data set used, the late afternoon dip in electron density observed near the F-peak and interpreted as the signature of the mid-latitude trough is well reproduced by the calculations. Calculations indicate that it does not arise from long residence times of plasma on the nightside, but is the signature of a gap between two major ionization sources, viz. photoionization and particle precipitation.

Non-Maxwellian ion velocity distributions observed using EISCAT

Recent observations from the EISCAT incoherent scatter radar have revealed bursts of poleward ion flow in the dayside auroral ionosphere which are consistent with the ionospheric signature of flux transfer events at the magnetopause. These bursts frequently contain ion drifts which exceed the neutral thermal speed and, because the neutral thermospheric wind is incapable of responding sufficiently rapidly, toroidal, non-Maxwellian ion velocity distributions are expected. The EISCAT observations are made with high time resolution (15 seconds) and at a large angle to the geomagnetic field (73.5°), allowing the non-Maxwellian nature of the distribution to be observed remotely for the first time. The observed features are also strongly suggestive of a toroidal distribution: characteristic spectral shape, increased scattered power (both consistent with reduced Landau damping and enhanced electric field fluctuations) and excessively high line-of-sight ion temperatures deduced if a Maxwellian distribution is assumed. These remote sensing observations allow the evolution of the distributions to be observed. They are found to be non-Maxwellian whenever the ion drift exceeds the neutral thermal speed, indicating that such distributions can exist over the time scale of the flow burst events (several minutes).

A survey of simultaneous observations of the high-latitude ionosphere and interplanetary magnetic field with EISCAT and AMPTE-UKS

This paper surveys the results of simultaneous observations by the EISCAT incoherent scatter radar and the AMPTE-UKS satellite, made during three periods in September and October 1984, when AMPTE-UKS was in the solar wind on the dayside of the Earth and the UK-POLAR EISCAT experiment was measuring ionospheric parameters at invariant latitudes 70.8–75.0°. A total of 42 h of EISCAT convection velocity data, with 2.5 min resolution, were obtained, together with 28 h of simultaneous 5 s resolution AMPTE-UKS observations of the solar wind and interplanetary magnetic field (IMF). The general features of the AMPTE-UKS data are described in Section 2 and those of the EISCAT data are described in Sections 3 and 4. The main subjects discussed are the form of the plasma convection patterns and their dependence on all three components of the IMF (Section 5), the ionospheric response to abrupt changes in the IMF (Section 6), in particular a sharp ‘southward turning’ of the IMF on 27 October 1984, and a crossing of an IMF sector boundary. Section 7 describes ‘short lived rapid flow burst’, which are believed to be related to flux transfer events at the magnetopause.

The effects of explicit versus parameterized convection on the MJO in a large-domain high-resolution tropical case study. Part II: Processes leading to differences in MJO development

High-resolution simulations over a large tropical domain (∼20◦S–20◦N and 42◦E–180◦E) using both explicit and parameterized convection are analyzed and compared during a 10-day case study of an active Madden-Julian Oscillation (MJO) event. In Part II, the moisture budgets and moist entropy budgets are analyzed. Vertical subgrid diabatic heating profiles and vertical velocity profiles are also compared; these are related to the horizontal and vertical advective components of the moist entropy budget which contribute to gross moist stability, GMS, and normalized GMS (NGMS). The 4-km model with explicit convection and good MJO performance has a vertical heating structure that increases with height in the lower troposphere in regions of strong convection (like observations), whereas the 12-km model with parameterized convection and a poor MJO does not show this relationship. The 4-km explicit convection model also has a more top-heavy heating profile for the troposphere as a whole near and to the west of the active MJO-related convection, unlike the 12-km parameterized convection model. The dependence of entropy advection components on moisture convergence is fairly weak in all models, and differences between models are not always related to MJO performance, making comparisons to previous work somewhat inconclusive. However, models with relatively good MJO strength and propagation have a slightly larger increase of the vertical advective component with increasing moisture convergence, and their NGMS vertical terms have more variability in time and longitude, with total NGMS that is comparatively larger to the west and smaller to the east.

Convective updraught evaluation in high-resolution NWP simulations using single-Doppler radar measurements

This study presents an evaluation of the size and strength of convective updraughts in high-resolution simulations by the UK Met Office Unified Model (UM). Updraught velocities have been estimated from range–height indicator (RHI) Doppler velocity measurements using the Chilbolton advanced meteorological radar, as part of the Dynamical and Microphysical Evolution of Convective Storms (DYMECS) project. Based on mass continuity and the vertical integration of the observed radial convergence, vertical velocities tend to be underestimated for convective clouds due to the undetected cross-radial convergence. Velocity fields from the UM at a resolution corresponding to the radar observations are used to scale such estimates to mitigate the inherent biases. The analysis of more than 100 observed and simulated storms indicates that the horizontal scale of updraughts in simulations tend to decrease with grid length; the 200 m grid length agreed most closely with the observations. Typical updraught mass fluxes in the 500 m grid length simulations were up to an order of magnitude greater than observed, and greater still in the 1.5 km grid length simulations. The effect of increasing the mixing length in the sub-grid turbulence scheme depends on the grid length. For the 1.5 km simulations, updraughts were weakened though their horizontal scale remained largely unchanged. Progressively more so for the sub-kilometre grid lengths, updraughts were broadened and intensified; horizontal scale was now determined by the mixing length rather than the grid length. In general, simulated updraughts were found to weaken too quickly with height. The findings were supported by the analysis of the widths of reflectivity patterns in both the simulations and observations.