A hardware design of a massive-parallel, modular NN-based vector quantizer for real-time video coding

This report describes the design of a modular, massive-parallel, neural-network (NN)-based vector quantizer for real-time video coding. The NN is a self-organizing map (SOM) that works only in the training phase for codebook generation, only at the recall phase for real-time image coding, or in both phases for adaptive applications. The neural net can be learned using batch or adaptive training and is controlled by an inside circuit, finite-state machine-based hard controller. The SOM is described in VHDL and implemented on electrically (FPGA) and mask (standard-cell) programmable devices.     

Beacon-Less Geographic Routing in Real Wireless Sensor Networks

Geographic Routing(GR)algorithms require nodes to periodically transmit HELLO messages to allow neigh- bors to know their positions(beaconing mechanism).Beacon-less routing algorithms have recently been proposed to reduce the control overheads due to these messages.However,existing beacon-less algorithms have not considered realistic physical layers.Therefore,those algorithms cannot work properly in realistic scenarios.In this paper we present a new beacon- less routing protocol called BOSS.Its design is based on the conclusions of our open-field experiments using Tmote-sky sensors.BOSS is adapted to error-prone networks and incorporates a new mechanism to reduce collisions and duplicate messages produced during the selection of the next forwarder node.We compare BOSS with Beacon-Less Routing(BLR) and Contention-Based Forwarding(CBF)algorithms through extensive simulations.The results show that our scheme is able to achieve almost perfect packet delivery ratio(like BLR)while having a low bandwidth consumption(even lower than CBF).Additionally,we carried out an empirical evaluation in a real testbed that shows the correctness of our simulation results.     

Molecular modeling of tricyclic compounds with anilino substituents and their intercalation complexes with DNA sequences

Although 9-anilinoacridines are among the best studied antitumoral intercalators, there are few studies about the effect of isosteric substitution of a benzene moiety for a heterocycle ring in the acridine framework. According to these studies, this approach may lead to effective cytotoxic agents, but good cytotoxic activity depends on structural requirements in the aniline ring which differ from those in 9-anilinoacridines. The present paper deals with molecular modeling studies of some 9-anilino substituted tricyclic compounds and their intercalation complexes (in various DNA sequences) resulting from docking the compounds into various DNA sequences. As expected, the isosteric substitution in 9-anilinoacridines influences the LUMO energy values and orbital distribution, the dipole moment, electrostatic charges and the conformation of the anilino ring. Other important differences are observed during the docking studies, for example, changes in the spatial arrangement of the tricyclic nucleus and the anilino ring at the intercalation site. Semiempirical calculations of the intercalation complexes show that the isosteric replacement of a benzene ring in the acridine nucleus affects not only DNA affinity but also base pair selectivity. These findings explain, at least partially, the different structural requirements observed in several 9-anilino substituted tricyclic compounds for cytotoxic activity. Thus, the data presented here may guide the rational design of new agents with different DNA binding properties and/or a cytotoxic profile by isosteric substitution of known intercalators.     

Adaptation and application of multi-objective evolutionary algorithms for rule reduction and parameter tuning of fuzzy rule-based systems

Recently, multi-objective evolutionary algorithms have been applied to improve the difficult tradeoff between interpretability and accuracy of fuzzy rule-based systems. It is known that both requirements are usually contradictory, however, these kinds of algorithms can obtain a set of solutions with different trade-offs. This contribution analyzes different application alternatives in order to attain the desired accuracy/interpr-etability balance by maintaining the improved accuracy that a tuning of membership functions could give but trying to obtain more compact models. In this way, we propose the use of multi-objective evolutionary algorithms as a tool to get almost one improved solution with respect to a classic single objective approach (a solution that could dominate the one obtained by such algorithm in terms of the system error and number of rules). To do that, this work presents and analyzes the application of six different multi-objective evolutionary algorithms to obtain simpler and still accurate linguistic fuzzy models by performing rule selection and a tuning of the membership functions. The results on two different scenarios show that the use of expert knowledge in the algorithm design process significantly improves the search ability of these algorithms and that they are able to improve both objectives together, obtaining more accurate and at the same time simpler models with respect to the single objective based approach.

---

Control of 1-D parabolic PDEs with Volterra nonlinearities, Part I: Design

Boundary control of nonlinear parabolic PDEs is an open problem with applications that include fluids, thermal, chemically-reacting, and plasma systems. In this paper we present stabilizing control designs for a broad class of nonlinear parabolic PDEs in 1-D. Our approach is a direct infinite dimensional extension of the finite-dimensional feedback linearization/backstepping approaches and employs spatial Volterra series nonlinear operators both in the transformation to a stable linear PDE and in the feedback law. The control law design consists of solving a recursive sequence of linear hyperbolic PDEs for the gain kernels of the spatial Volterra nonlinear control operator. These PDEs evolve on domains T_n of increasing dimensions n+1 and with a domain shape in the form of a “hyper-pyramid”, 0≤ξ_n≤ξ_n−1?≤ξ₁≤x≤1. We illustrate our design method with several examples. One of the examples is analytical, while in the remaining two examples the controller is numerically approximated. For all the examples we include simulations, showing blow up in open loop, and stabilization for large initial conditions in closed loop. In a companion paper we give a theoretical study of the properties of the transformation, showing global convergence of the transformation and of the control law nonlinear Volterra operators, and explicitly constructing the inverse of the feedback linearizing Volterra transformation; this, in turn, allows us to prove L² and H¹ local exponential stability (with an estimate of the region of attraction where possible) and explicitly construct the exponentially decaying closed loop solutions.     

Prediction of the percentages of cows'', goats'' and ewes'' milk in “Iberico” cheese by electrophoretic analysis of whey proteins

Stepwise multiple linear regression (SMLR) and principal components regression (PCR) have been used to predict the percentages of cows', goats' and ewes' milk in Iberico cheese, using the results obtained by electrophoretic analysis (PAGE and IEF) of whey proteins, using standard cheeses. Similar predictions of the percentages of milks from the three species were obtained when either SMLR or PCR were applied to the electrophoretic data, i.e. the optical intensity of the electrophoretic bands (PAGE or IEF) of the whey proteins. The root mean square error of prediction in cross-validation (RMSEPCV) was lower than 4% in all cases.     

Interpretation time of serial chest CT examinations with stacked-metaphor workstation versus film alternator

PURPOSE: Interpretation time of serial staging chest CT cases, which each contained current and previous examinations, with a simple prototype workstation called filmstack was experimentally compared with interpretation time with a film alternator. MATERIALS AND METHODS: The filmstack displayed a "stack" of sections for each examination; user controls allowed rapid selection of preset attenuation windows and both synchronized and unsynchronized scrolling. Eight radiologists were timed as they used the filmstack and the film alternator to interpret four ergonomically complex serial CT cases. RESULTS: All reports dictated on the basis of findings with filmstack and film were of acceptable clinical accuracy. The time to examine a case with filmstack was significantly faster than the time with film, including the time to load and unload the alternator (99% confidence [P = .01]). There was no statistically significant difference in interpretation time between filmstack and prehung film. CONCLUSION: Use of a low-cost stacked CT workstation with a single 1,024 x 1,024 monitor is an effective means of interpreting cases that require comparison of multiple CT examinations.     

Low-contrast lesion segmentation in advanced MRI experiments by time-domain Ricker-type wavelets and fuzzy 2-means

Automated suspicious region segmentation has become a crucial need for the experts dealing with numerous images containing contrast-based lesions in MRI. Not all solutions, however, are based on mathematical infrastructure or providing adequate flexibility. On the other hand, segmentation of low-contrast lesions is very challenging for researchers; therefore, advanced magnetic resonance imaging (MRI) experiments are not commonly used in researches. Given the need of repeatability and adaptability, we present an automated framework for intelligent segmentation of brain lesions by wavelet imaging and fuzzy 2-means. Besides the general use of the wavelets in image processing, which is edge detection; we employed the second-order Ricker-type wavelets as the core of our novel imaging framework for low-contrast lesion segmentation. We firstly introduced the mathematical basis of several Ricker wavelet functions, which are in symmetrical form satisfying finite-energy and admissibility conditions of mother wavelets. Afterwards, we investigated three types of Ricker wavelets to apply on our clinical dataset containing susceptibility-weighted (SW) and minimum intensity projection SW (mIP-SW) images with barely-visible lesions. Finally, we adjusted the system parameters of the wavelets for optimization and post-segmentation by fuzzy 2-means. According to the preliminary results of the clinical experiments we conducted, our framework provided 93.53% average dice score (DSC) for SWI by Ricker-3 and 92.56% for mIP-SWI by Ricker-2 wavelet, as the main performance criteria of segmentation. Despite the lack of SWI or mIP-SWI experiments in the public datasets, we tested our framework with BraTS 2012 training sets containing real images with visible lesions and achieved an average of 88.13% DSC with 11.66% standard deviation by re-optimized framework for whole lesion segmentation, which is one of the highest among other relevant researches. In detail, 87.52% DSC for LG datasets with 11.32% standard deviation; while 88.34% DSC for HG datasets with 11.77% standard deviation are calculated.

     

Using dynamic demand information and zoning for the storage of non-uniform density stock keeping units

The warehouse order-picking operation is one of the most labour-intense activities that has an important impact on responsiveness and efficiency of the supply chain. An understanding of the impact of the simultaneous effects of customer demand patterns and order clustering, considering physical restrictions in product storage, is critical for improving operational performance. Storage restrictions may include storing non-uniform density stock keeping units (SKUs) whose dimensions and weight constrain the order-picking operation given that a priority must be followed. In this paper, a heuristic optimisation based on a quadratic integer programming is employed to generate a layout solution that considers customer demand patterns and order clustering. A simulation model is used to investigate the effects of creating and implementing these layout solutions in conjunction with density zones to account for restrictions in non-uniform density SKUs. Results from combining layout optimisation heuristics and density zoning indicate statistical significant differences between assignments that ignore the aforementioned factors and those that recognise it.