The influence of ARIMA-GARCH parameters in feed forward neural networks prediction

The objective of this article is to find out the influence of the parameters of the ARIMA-GARCH models in the prediction of artificial neural networks (ANN) of the feed forward type, trained with the Levenberg–Marquardt algorithm, through Monte Carlo simulations. The paper presents a study of the relationship between ANN performance and ARIMA-GARCH model parameters, i.e. the fact that depending on the stationarity and other parameters of the time series, the ANN structure should be selected differently. Neural networks have been widely used to predict time series and their capacity for dealing with non-linearities is a normally outstanding advantage. However, the values of the parameters of the models of generalized autoregressive conditional heteroscedasticity have an influence on ANN prediction performance. The combination of the values of the GARCH parameters with the ARIMA autoregressive terms also implies in ANN performance variation. Combining the parameters of the ARIMA-GARCH models and changing the ANN’s topologies, we used the Theil inequality coefficient to measure the prediction of the feed forward ANN.     

Shape-sensitive MLS deformation

This work presents methods for deforming meshes in a shape-sensitive way using Moving Least Squares (MLS) optimization. It extends an approach for deforming space (Cuno et al. in Proceedings of the 27th Computer Graphics International Conference, pp. 115–122, 2007) by showing how custom distance metrics may be used to achieve deformations which preserve the overall mesh shape. Several variant formulations are discussed and demonstrated, including the use of geodesic distances, distances constrained to paths contained in the mesh, the use of skeletons, and a reformulation of the MLS scheme which makes it possible to affect the bending behavior of the deformation. Finally, aspects of the implementation of these techniques in parallel architectures such as GPUs (graphics processing units) are described and compared with CPU-only implementations.     

Realistic real-time sound re-synthesis and processing for interactive virtual worlds

We present new GPU-based techniques for implementing linear digital filters for real-time audio processing. Our solution for recursive filters is the first presented in the literature. We demonstrate the relevance of these algorithms to computer graphics by synthesizing realistic sounds of colliding objects made of different materials, such as glass, plastic, and wood, in real time. The synthesized sounds can be parameterized by the object materials, velocities, and collision angles. Despite its flexibility, our approach uses very little memory, since it essentially requires a set of coefficients representing the impulse response of each material sound. Such features make our approach an attractive alternative to traditional CPU-based techniques that use playback of pre-recorded sounds.     

Hand gesture recognition from depth and infrared Kinect data for CAVE applications interaction

This paper presents a real-time framework that combines depth data and infrared laser speckle pattern (ILSP) images, captured from a Kinect device, for static hand gesture recognition to interact with CAVE applications. At the startup of the system, background removal and hand position detection are performed using only the depth map. After that, tracking is started using the hand positions of the previous frames in order to seek for the hand centroid of the current one. The obtained point is used as a seed for a region growing algorithm to perform hand segmentation in the depth map. The result is a mask that will be used for hand segmentation in the ILSP frame sequence. Next, we apply motion restrictions for gesture spotting in order to mark each image as a ‘Gesture’ or ‘Non-Gesture’. The ILSP counterparts of the frames labeled as “Gesture” are enhanced by using mask subtraction, contrast stretching, median filter, and histogram equalization. The result is used as the input for the feature extraction using a scale invariant feature transform algorithm (SIFT), bag-of-visual-words construction and classification through a multi-class support vector machine (SVM) classifier. Finally, we build a grammar based on the hand gesture classes to convert the classification results in control commands for the CAVE application. The performed tests and comparisons show that the implemented plugin is an efficient solution. We achieve state-of-the-art recognition accuracy as well as efficient object manipulation in a virtual scene visualized in the CAVE.     

Position USBL/DVL sensor-based navigation filter in the presence of unknown ocean currents

This paper presents a novel approach to the design of globally asymptotically stable (GAS) position and velocity filters for Autonomous Underwater Vehicles (AUVs) based directly on the sensor readings of an Ultra-short Baseline (USBL) acoustic array system and a Doppler Velocity Log (DVL). The proposed methodology is based on an equivalent linear time-varying (LTV) system that fully captures the dynamics of the nonlinear system, allowing for the use of powerful linear system analysis and filtering design tools that yield GAS filter error dynamics. Numerical results using Monte Carlo simulations and comparison to the Bayesian Cramér Rao Bound (BCRB) reveal that the performance of the proposed filter is tight to this theoretical estimation error lower bound. In comparison with other approaches, the present solution achieves the same level of performance of the Extended Kalman Filter (EKF), which does not offer GAS guarantees, and outperforms other classical filtering approaches designed in inertial coordinates instead of the body-fixed coordinate frame.     

Persistent Scatterers Interferometry detects and measures ground subsidence in Lisbon

Results from the application of Persistent Scatterers Interferometry in Lisbon Metropolitan Area revealed two previously unknown subsiding urban areas: one (Laranjeiras) is located in the center of Lisbon; another (Vialonga) is to be found toward the North, in an industrial region crossed by Lisbon's main highway and railway lines. The two subsiding sectors are bordered by sharp velocity gradients, and the subsidence pattern appears partially delimited by mapped geologic faults. Surface geology and urbanization alone are unable to explain the phenomena. In the Vialonga area, the historical record of water pore pressure shows a clear decline of the levels (up to 65 m in 27 years), providing evidence of over-exploitation of groundwater resources. Limited information from wells drilled inside and outside the subsidence area points to a spatial correlation between the subsidence and the water pressure levels, and suggests that faults could be acting as hydraulic barriers in the aquifer system. The surface subsidence detected is probably caused by compaction of a clay-rich Oligocene-aged aquitard, led by over-exploitation of adjacent aquifers. The same Oligocene aquitard layer is present in the Laranjeiras area, immediately bellow a multi-layered sand-clay-limestone Miocene aquifer, but further work is needed to diagnose the possibility of over-exploitation of groundwater here. In this work we were able to independently confirm the PSI results, by comparing autonomous PSI results processed for the same geographical areas, and by comparing PSI with leveling and continuous GPS derived subsidence velocities, whose close match provided further ground validation of the space-borne PSI technique.     

TinyPBC: Pairings for authenticated identity-based non-interactive key distribution in sensor networks

Key distribution in Wireless Sensor Networks (WSNs) is challenging. Symmetric cryptosystems can perform it efficiently, but they often do not provide a perfect trade-off between resilience and storage. Further, even though conventional public key and elliptic curve cryptosystems are computationally feasible on sensor nodes, protocols based on them are not, as they require the exchange and storage of large keys and certificates, which is expensive.Using Pairing-Based Cryptography (PBC) protocols parties can agree on keys without any interaction. In this work, we (i) show how security in WSNs can be bootstrapped using an authenticated identity-based non-interactive protocol and (ii) present TinyPBC, to our knowledge, the most efficient implementation of PBC primitives for 8, 16 and 32-bit processors commonly found in sensor nodes. TinyPBC is able to compute pairings, the most expensive primitive of PBC, in 1.90 s on ATmega128L, 1.27 s on MSP430 and 0.14 s on PXA27x.     

Simulations of extensional flow in microrheometric devices

We present a detailed numerical study of the flow of a Newtonian fluid through microrheometric devices featuring a sudden contraction–expansion. This flow configuration is typically used to generate extensional deformations and high strain rates. The excess pressure drop resulting from the converging and diverging flow is an important dynamic measure to quantify if the device is intended to be used as a microfluidic extensional rheometer. To explore this idea, we examine the effect of the contraction length, aspect ratio and Reynolds number on the flow kinematics and resulting pressure field. Analysis of the computed velocity and pressure fields show that, for typical experimental conditions used in microfluidic devices, the steady flow is highly three-dimensional with open spiraling vortical structures in the stagnant corner regions. The numerical simulations of the local kinematics and global pressure drop are in good agreement with experimental results. The device aspect ratio is shown to have a strong impact on the flow and consequently on the excess pressure drop, which is quantified in terms of the dimensionless Couette and Bagley correction factors. We suggest an approach for calculating the Bagley correction which may be especially appropriate for planar microchannels. Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorized users.     

Digital Butterworth filter for subtracting noise from low magnitude surface electromyogram

This work presents a digital filter designed to delimitate the frequency band of surface electromyograms (EMG) and remove the mains noise and its harmonics, focusing the signal analysis during reduced muscle activity. A Butterworth filter was designed as the frequency-domain product of a second order, high-pass filter with cutoff frequency 10 Hz, an eighth order low-pass filter, with cutoff at 400 Hz and six stop-band filters, second order, centered at the 60 Hz mains noise and its harmonics until 360 Hz. The resulting filter was applied in both direct and reverse directions of the signals to avoid phase distortions. The performance was evaluated with a simulated EMG signal with additive noise in multiples of 60 Hz. A qualitative assessment was made with real EMG data, acquired from 16 subjects, with age from 20 to 32 years. Subjects were positioned in orthostatic position during 21s, being only the last second analyzed to assure stationarity. EMG were collected by Ag/AgCl electrodes on right lateral gastrocnemius, amplified with gain 5000, filtered in the band from 10 Hz to 1 kHz, and thus digitized with 2ksamples/s. The filter effectively removed the mains noise components, with attenuations greater than 96.6%. The attenuation of the simulated signal at frequencies below 15 Hz and at 60 Hz caused only a small reduction of total power, preserving the original spectrum. Thus, the filter resulted suitable to the proposed application.