A digital architecture for support vector machines: theory, algorithm, and FPGA implementation

In this paper, we propose a digital architecture for support vector machine (SVM) learning and discuss its implementation on a field programmable gate array (FPGA). We analyze briefly the quantization effects on the performance of the SVM in classification problems to show its robustness, in the feedforward phase, respect to fixed-point math implementations; then, we address the problem of SVM learning. The architecture described here makes use of a new algorithm for SVM learning which is less sensitive to quantization errors respect to the solution appeared so far in the literature. The algorithm is composed of two parts: the first one exploits a recurrent network for finding the parameters of the SVM; the second one uses a bisection process for computing the threshold. The architecture implementing the algorithm is described in detail and mapped on a real current-generation FPGA (Xilinx Virtex II). Its effectiveness is then tested on a channel equalization problem, where real-time performances are of paramount importance.     

Transmission Electron Microscopy and Electron Energy-Loss Spectroscopy Characterization of Glass Phase in Sol-Gel-Derived Mullite

Sol-gel-derived mullite ceramics were processed by pressureless sintering at 1600°, 1650°, and 1700°C for 4 h. Microstructural and microchemical characterization of the mullite materials was performed using transmission electron microscopy, in conjunction with energy-dispersive X-ray spectroscopy and electron energy-loss spectroscopy (EELS). Apart from mullite grain diameter and triplepocket size, no major microstructural changes were observed with increasing sintering temperature. Residual glass was present at triple pockets and along two-grain junctions. Not all grain boundaries revealed the presence of a continuous amorphous intergranular film. Clean interfaces were observed only at boundaries with one grain parallel to the [001] orientation (low-energy configuration). Quantitative EELS analysis of mullite grains and glass pockets revealed only small changes in composition with increasing sintering temperature; i.e., the alumina:silica ratio slightly increased for mullite and glass. The analysis implied that mullite with this relatively high aluminum content would not be stable adjacent to residual glass. However, a stable glass-mullite system has been proposed, because impurity cations were detected within glass pockets, which suggested a slight shift of the subsolidus line (glass-mullite/ mullite) to a higher amount of alumina. Energy-loss nearedge structure studies of the Si- L _2,3 edge revealed a similar near-edge structure for the mullite, residual glass, and quartz. Thus, SiO₄ tetrahedra were thought to be the main building units of the glass contained in sintered mullite.     

Data collection in global software engineering research: learning from past experience

Global Software Engineering has become a standard in today’s software industry. Research in distributed software development poses severe challenges that are due to the spatial and temporal distribution of the actors, as well as to language, intercultural and organizational aspects. These challenges occur in addition to “traditional” challenges of the domain itself in large-scale software projects, like coordination and communication issues, requirements volatily, lack of domain knowledge, among others. While several authors have reported empirical studies of global software development projects, the methodological difficulties and challenges of this type of studies have not been sufficiently discussed. In this paper, we share our experiences of collecting and analysing qualitative data in the context of Global Software Engineering projects. We discuss strategies for gaining access to field sites, building trust and documenting distributed and complex work practices in the context of several research projects we have conducted in the past 9 years. The experiences described in this paper illustrate the need to deal with fundamental problems, such as understanding local languages and different cultures, observing synchronous interaction, or dealing with barriers imposed by political conflicts between the sites. Based on our findings, we discuss some practical implications and strategies that can be used by other researchers and provide some recommendations for future research in methodological aspects of Global Software Engineering.     

The role of lanthana loading on the catalytic properties of Pd/Al2O3-La2O3 in the NO reduction with H2

The role of La₂O₃ loading in Pd/Al₂O₃-La₂O₃ prepared by sol–gel on the catalytic properties in the NO reduction with H₂ was studied. The catalysts were characterized by N₂ physisorption, temperature-programmed reduction, differential thermal analysis, temperature-programmed oxidation and temperature-programmed desorption of NO.

The physicochemical properties of Pd catalysts as well as the catalytic activity and selectivity are modified by La₂O₃ inclusion. The selectivity depends on the NO/H₂ molar ratio (GHSV = 72,000 h⁻¹) and the extent of interaction between Pd and La₂O₃. At NO/H₂ = 0.5, the catalysts show high N₂ selectivity (60–75%) at temperatures lower than 250 °C. For NO/H₂ = 1, the N₂ selectivity is almost 100% mainly for high temperatures, and even in the presence of 10% H₂O vapor. The high N₂ selectivity indicates a high capability of the catalysts to dissociate NO upon adsorption. This property is attributed to the creation of new adsorption sites through the formation of a surface PdO_x phase interacting with La₂O₃. The formation of this phase is favored by the spreading of PdO promoted by La₂O₃. DTA shows that the phase transformation takes place at temperatures of 280–350 °C, while TPO indicates that this phase transformation is related to the oxidation process of PdO: in the case of Pd/Al₂O₃ the O₂ uptake is consistent with the oxidation of PdO to PdO₂, and when La₂O₃ is present the O₂ uptake exceeds that amount (1.5 times). La₂O₃ in Pd catalysts promotes also the oxidation of Pd and dissociative adsorption of NO mainly at low temperatures (<250 °C) favoring the formation of N₂.     

Experimental design to evaluate properties of electrospun fibers of zein/poly (ethylene oxide) for biomaterial applications

The production of polymer fibers from the combination of zein and PEO might have great potential in the field of biomaterial. Zein/PEO fibers were obtained in this work through solution electrospinning. An experimental design, 2^4-1, was used for evaluating the influences of PEO content in the blend, distance from the needle tip to the collector, applied electric voltage and solution flow for average fiber diameter and relative-yield process. Beyond this, the relationship between PEO content in the blend and the fiber properties were evaluated through FTIR, DSC, TG, tensile tests, and cytotoxic tests. The factor that exerts the greatest effect on the average fiber diameter response was the electrical voltage. The increase in PEO content in the blend decreased the thermal stability and increased the degree of the fibers' crystallinity. The mechanical tests showed that fibers with higher elongation were obtained at richer PEO blends. The fibers presented cytocompatible characteristics.     

SONIC: streaming overlapping community detection

A community within a graph can be broadly defined as a set of vertices that exhibit high cohesiveness (relatively high number of edges within the set) and low conductance (relatively low number of edges leaving the set). Community detection is a fundamental graph processing analytic that can be applied to several application domains, including social networks. In this context, communities are often overlapping, as a person can be involved in more than one community (e.g., friends, and family); and evolving, since the structure of the network changes. We address the problem of streaming overlapping community detection, where the goal is to maintain communities in the presence of streaming updates. This way, the communities can be updated more efficiently. To this end, we introduce SONIC—a find-and-merge type of community detection algorithm that can efficiently handle streaming updates. SONIC first detects when graph updates yield significant community changes. Upon the detection, it updates the communities via an incremental merge procedure. The SONIC algorithm incorporates two additional techniques to speed-up the incremental merge; min-hashing and inverted indexes. Results show that SONIC can provide high quality overlapping communities, while handling streaming updates several orders of magnitude faster than the alternatives performing from-scratch computation.     

Validation of static properties in unified modeling language models for cyber physical systems

Cyber physical systems (CPSs) can be found nowadays in various fields of activity. The increased interest for these systems as evidenced by the large number of applications led to complex research regarding the most suitable methods for design and development. A promising solution for specification, visualization, and documentation of CPSs uses the Object Management Group (OMG) unified modeling language (UML). UML models allow an intuitive approach for embedded systems design, helping end-users to specify the requirements. However, the UML models are represented in an informal language. Therefore, it is difficult to verify the correctness and completeness of a system design. The object constraint language (OCL) was defined to add constraints to UML, but it is deficient in strict notations of mathematics and logic that permits rigorous analysis and reasoning about the specifications. In this paper, we investigated how CPS applications modeled using UML deployment diagrams could be formally expressed and verified. We used Z language constructs and prototype verification system (PVS) as formal verification tools. Considering some relevant case studies presented in the literature, we investigated the opportunity of using this approach for validation of static properties in CPS UML models.