Learning eigenfunctions links spectral embedding and kernel PCA

In this letter, we show a direct relation between spectral embedding methods and kernel principal components analysis and how both are special cases of a more general learning problem: learning the principal eigenfunctions of an operator defined from a kernel and the unknown data-generating density. Whereas spectral embedding methods provided only coordinates for the training points, the analysis justifies a simple extension to out-of-sample examples (the Nystr?m formula) for multidimensional scaling (MDS), spectral clustering, Laplacian eigenmaps, locally linear embedding (LLE), and Isomap. The analysis provides, for all such spectral embedding methods, the definition of a loss function, whose empirical average is minimized by the traditional algorithms. The asymptotic expected value of that loss defines a generalization performance and clarifies what these algorithms are trying to learn. Experiments with LLE, Isomap, spectral clustering, and MDS show that this out-of-sample embedding formula generalizes well, with a level of error comparable to the effect of small perturbations of the training set on the embedding.     

Semiautomated high-throughput extraction and cleanup method for the measurement of polybrominated diphenyl ethers and polybrominated and polychlorinated biphenyls in breast milk

A semiautomated extraction and cleanup method has been developed to measure eight polybrominated diphenyl ethers (PBDEs), 2,2',4,4',5,5'-hexabromobiphenyl (BB-153), and 2,2',4,4',5,5'-hexachlorobiphenyl (CB-153). The method employs solid-phase dispersion on diatomaceous earth in a solid-phase extraction cartridge followed by automated addition of internal standards ((13)C-labeled). Extraction is then performed using an automated modular solid-phase extraction system. The extraction procedure includes drying the sample on diatomaceous earth by pressurized nitrogen and eluting target analytes and lipids with dichloromethane. Lipid content is determined gravimetrically. Lipid determinations performed using this method are compared with other standard methods and with a certified reference material. A relative standard deviation of 7.9% was obtained for 130 determinations of the lipid content in a breast milk quality control sample. Final analytical determination of target analytes was performed by gas chromatography-isotope dilution high-resolution mass spectrometry. Relative standard deviations for the measurements of target analytes for which a labeled internal standard was available were below 10% for analytes at concentrations above 1 ng/g of lipid. Mean recoveries of the (13)C-labeled internal standards ranged from 60 to 89% for the eight PBDE congeners; 74 and 113% were recovered for BB-153 and CB-153, respectively.     

Study of the oxidation kinetics of finely-divided magnetites. II - Influence of chromium substitution

The oxidation kinetics of magnetites substituted by chromium (Fe²⁺Fe³⁺_2?xCr³⁺_x)O^2?₄ (0 < x < 2) into γ(Fe³⁺_1?yCr³⁺_y)₂O^2?₃, (x = 3y), which is a metastable phase, was found out to be ruled by the law of diffusion, under variable working conditions, of vacancies generated at the surface. The chemical diffusion coefficient is a function of the substitution ratio, crystallite size and the number of vacancies in the spinel lattice. Contrary to magnetites substituted by aluminum, the activation energy varies irregularly with the substitution ratio.     

Ultrasonic techniques for imaging and measurements in molten aluminum

In order to achieve net shape forming, processing of aluminum (Al) in the molten state is often necessary. However, few sensors and techniques have been reported in the literature due to difficulties associated with molten Al, such as high temperature, corrosiveness, and opaqueness. In this paper, development of ultrasonic techniques for imaging and measurements in molten Al using buffer rods operated at 10 MHz is presented. The probing end of the buffer rod, having a flat surface or an ultrasonic lens, was immersed into molten Al while the other end with an ultrasonic transducer was air-cooled to room temperature. An ultrasonic image of a character "N", engraved on a stainless steel plate immersed in molten Al, and its corrosion have been observed at 780 degrees C using the focused probe in ultrasonic pulse-echo mode. Because cleanliness of molten Al is crucial for part manufacturing and recycling in Al processing, inclusion detection experiments also were carried out using the nonfocused probe in pitch-catch and pulse-echo modes. Backscattered ultrasonic signals from manually added silicon carbide particles, with an average diameter of 50 microm, in molten Al have been successfully observed at 780 degrees C. For optimal image quality, the spatial resolution of the focused probe was crucial, and the high signal-to-noise ratio of the nonfocused probe was the prime factor responsible for the inclusion detection sensitivity using backscattered ultrasonic signals. In addition, it was found that ultrasound could provide an alternative method for evaluating the degree of wetting between a solid material and a molten metal. Our experimental results showed that there was no ultrasonic coupling at the interface between an alumina rod and molten Al up to 1000 degrees C; therefore, no wetting existed at this interface. Also because ultrasonic velocity in alumina is temperature dependent, this rod proved to be able to be used as an in-line temperature monitoring sensor under 1000 degrees C in molten Al.     

Cadmium and lead adsorption by a natural polysaccharide in MF membrane reactor: experimental analysis and modelling

In the present work, Pb(2+) and Cd(2+) adsorption onto a natural polysaccharide has been studied in membrane reactors. The process involves a stirred semi-batch reactor for the adsorption step and a microfiltration (MF) process in order to confine the particles. Due to their lower affinity for the biosorbent, Cd(2+) ions were found to breakthrough the process faster than Pb(2+) cations. The experimental results showed the technical feasibility of the pilot. A mass balance model based on the Langmuir equilibrium isotherm was used to describe the adsorption process. This relation is able to predict experimental data under different operating conditions: the adsorbent and metal concentrations, and the permeate flow rate. Based on these results, it is demonstrated that the biosorbent studied represents an interesting low-cost solution for the treatment of metal ions polluted waters.     

Free-Sets: A Condensed Representation of Boolean Data for the Approximation of Frequency Queries

Given a large collection of transactions containing items, a basic common data mining problem is to extract the so-called frequent itemsets (i.e., sets of items appearing in at least a given number of transactions). In this paper, we propose a structure called free-sets, from which we can approximate any itemset support (i.e., the number of transactions containing the itemset) and we formalize this notion in the framework of -adequate representations (H. Mannila and H. Toivonen, 1996. In Proc. of the Second International Conference on Knowledge Discovery and Data Mining (KDD'96), pp. 189–194). We show that frequent free-sets can be efficiently extracted using pruning strategies developed for frequent itemset discovery, and that they can be used to approximate the support of any frequent itemset. Experiments on real dense data sets show a significant reduction of the size of the output when compared with standard frequent itemset extraction. Furthermore, the experiments show that the extraction of frequent free-sets is still possible when the extraction of frequent itemsets becomes intractable, and that the supports of the frequent free-sets can be used to approximate very closely the supports of the frequent itemsets. Finally, we consider the effect of this approximation on association rules (a popular kind of patterns that can be derived from frequent itemsets) and show that the corresponding errors remain very low in practice.     

A tabu search heuristic for the quay crane scheduling problem

This paper proposes a tabu search heuristic for the Quay Crane Scheduling Problem (QCSP), the problem of scheduling a fixed number of quay cranes in order to load and unload containers into and from a ship. The optimality criterion considered is the minimum completion time. Precedence and non-simultaneity constraints between tasks are taken into account. The former originate from the different kind of operations that each crane has to perform; the latter are needed in order to avoid interferences between the cranes. The QCSP is decomposed into a routing problem and a scheduling problem. The routing problem is solved by a tabu search heuristic, while a local search technique is used to generate the solution of the scheduling problem. This is done by minimizing the longest path length in a disjunctive graph. The effectiveness of our algorithm is assessed by comparing it to a branch-and-cut algorithm and to a Greedy Randomized Adaptive Search Procedure (GRASP).     

Well-structured languages

This paper introduces the notion of well-structured language. A well-structured language can be defined by a labelled well-structured transition system, equipped with an upward-closed set of accepting states. That peculiar class of transition systems has been extensively studied in the field of computer-aided verification, where it has direct an important applications. Petri nets, and their monotonic extensions (like Petri nets with non-blocking arcs or Petri nets with transfer arcs), for instance, are special subclasses of well-structured transition systems. We show that the class of well-structured languages enjoy several important closure properties. We propose several pumping lemmata that are applicable respectively to the whole class of well-structured languages and to the classes of languages recognized by Petri nets or Petri nets with non-blocking arcs. These pumping lemmata allow us to characterize the limits in the expressiveness of these classes of language. Furthermore, we exploit the pumping lemmata to strictly separate the expressive power of Petri nets, Petri nets with non-blocking arcs and Petri nets with transfer arcs.     

System integration of high voltage electrostatic MEMS actuators

A system integration for High Voltage (HV) electrostatic MicroElectroMechanical Systems (MEMS) actuators is introduced on a micro-Printed Circuit Board. The system includes a programmable microcontroller, a programmable DC/DC converter, a multi output HV interface and electrostatic MEMS actuators. The system produces high output voltages (10–300 V) and can control a large variety of MEMS capacitive loads (1 to 50 pF) by combining diverse semiconductor technologies. This system proves that technologies, such as low voltage CMOS of different processes, high voltage DMOS and MEMS, can interact, communicate and even be integrated as a System In Package (SIP), providing significant size and cost reductions. The system was programmed to control electrostatic MEMS actuator. The DC/DC converter was made from components of different technologies and two addressable high voltage CMOS interfaces were fabricated with DALSA's 0.8 μm High Voltage process. A prototype of the global system has been built and tested.