Structured Output SVM for Remote Sensing Image Classification

Traditional kernel classifiers assume independence among the classification outputs. As a consequence, each misclassification receives the same weight in the loss function. Moreover, the kernel function only takes into account the similarity between input values and ignores possible relationships between the classes to be predicted. These assumptions are not consistent for most of real-life problems. In the particular case of remote sensing data, this is not a good assumption either. Segmentation of images acquired by airborne or satellite sensors is a very active field of research in which one tries to classify a pixel into a predefined set of classes of interest (e.g. water, grass, trees, etc.). In this situation, the classes share strong relationships, e.g. a tree is naturally (and spectrally) more similar to grass than to water. In this paper, we propose a first approach to remote sensing image classification using structured output learning. In our approach, the output space structure is encoded using a hierarchical tree, and these relations are added to the model in both the kernel and the loss function. The methodology gives rise to a set of new tools for structured classification, and generalizes the traditional non-structured classification methods. Comparison to standard SVM is done numerically, statistically and by visual inspection of the obtained classification maps. Good results are obtained in the challenging case of a multispectral image of very high spatial resolution acquired with QuickBird over a urban area.     

Interpenetrated Gel Polymer Binder for High‐Performance Silicon Anodes in Lithium‐ion Batteries

Silicon has attracted ever‐increasing attention as a high‐capacity anode material in Li‐ion batteries owing to its extremely high theoretical capacity. However, practical application of silicon anodes is seriously hindered by its fast capacity fading as a result of huge volume changes during the charge/discharge process. Here, an interpenetrated gel polymer binder for high‐performance silicon anodes is created through in‐situ crosslinking of water‐soluble poly(acrylic acid) (PAA) and polyvinyl alcohol (PVA) precursors. This gel polymer binder with deformable polymer network and strong adhesion on silicon particles can effectively accommodate the large volume change of silicon anodes upon lithiation/delithiation, leading to an excellent cycling stability and high Coulombic efficiency even at high current densities. Moreover, high areal capacity of ～4.3 mAh/cm² is achieved based on the silicon anode using the gel PAA–PVA polymer binder with a high mass loading. In view of simplicity in using the water soluble gel polymer binder, it is believed that this novel binder has a great potential to be used for high capacity silicon anodes in next generation Li‐ion batteries, as well as for other electrode materials with large volume change during cycling.     

Thermoelectric Properties of the Clathrate I Ba<Subscript>8</Subscript>Ge<Subscript>43</Subscript>□<Subscript>3</Subscript>

The clathrate I Ba₈Ge₄₃□₃ [space group Ia[`3]d Iabar{3}d , no. 230, a = 21.307(1) ?] has been synthesized as a single phase and characterized by x-ray powder diffraction and metallographic analysis. Electrical and thermal transport measurements have been performed in the temperature range of 5 K to 673 K. Ba₈Ge₄₃□₃ displays the electrical resistivity of a poor metal at low temperatures, with semiconducting-like behavior appearing above 300 K.     

2-Mercaptobenzothiazolate complexes of rare earth metals and their electroluminescent properties

We have demonstrated the electroluminescent (EL) properties of 2-mercaptobenzothiazolate complexes of rare earth metals [Ln(mbt)₃, Ln = Y, Sm, Eu, Gd, Tb, Dy, Tm] using simple non-doped two-layer organic light emitting diode with the configuration of indium tin oxide/N,N′-bis(3-methylphenyl)-N,N′-diphenylbenzidine/Ln(mbt)₃/Yb. It was found that 2-mercaptobenzothiazolate complexes have highly efficient intra-energy transfer from the singlet to the triplet state of the ligand, and then to the excited state of the central lanthanide ions. Thus Y(mbt)₃ and Gd(mbt)₃ exhibit the broad ligand-centered emission with maximum near 600 nm and Dy(mbt)₃, Tb(mbt)₃ and Tm(mbt)₃ complexes exhibit pure sharp emission bands from the intra f–f transitions of lanthanide ions Tb³⁺: ⁵D₄ → ⁷F₆ (492 nm), ⁵D₄ → ⁷F₅ (547 nm), ⁵D₄ → ⁷F₄ (589 nm), ⁵D₄ → ⁷F₃ (624 nm); Dy³⁺: ⁴F_9/2 → ⁶H_13/2 (575 nm) and Tm³⁺: ³H₄–³H₆ (795 нм).     

The effect of red pigment on the amyloidization of yeast proteins

The intensity of amyloid-bound thioflavine T fluorescence was studied in crude lysates of yeast strains carrying mutations in the ADE1 or ADE2 genes and accumulating the red pigment (a result of polymerization of aminoimidazoleribotide), and in white isogenic strains--either adenine prototrophs or carrying mutations at the first stages of purine biosynthesis. We found that the red pigment leads to a drop of amyloid content. This result, along with the data on separation of protein polymers of white and red strains in PAGE, suggests that the red pigment inhibits amyloid fibril formation. The differences in transmission of the thioflavine T fluorescence pattern by cytoduction and in blot-hybridization of pellet proteins of red and white [PSI(+) ] strains with Sup35p antibodies confirmed this conclusion. Purified red pigment treatment also led to a decrease of fluorescence intensity of thioflavine T bound to insulin fibrils and to yeast pellet protein aggregates from [PSI(+) ] strains. This suggests red pigment interaction with amyloid fibrils. Comparison of pellet proteins from red and white isogenic strains separated by 2D-electrophoresis followed by MALDI analysis has allowed us to identify 48 pigment-dependent proteins. These proteins mostly belong to functional classes of chaperones and proteins involved in glucose metabolism, closely corresponding to prion-dependent proteins that we characterized previously. Also present were some proteins involved in stress response and proteolysis. We suppose that the red pigment acts by blocking certain sites on amyloid fibrils that, in some cases, can lead in vivo to interfere with their contacts with chaperones and the generation of prion seeds.     

Peptides before and during the nucleotide world: an origins story emphasizing cooperation between proteins and nucleic acids

Recent developments in Origins of Life research have focused on substantiating the narrative of an abiotic emergence of nucleic acids from organic molecules of low molecular weight, a paradigm that typically sidelines the roles of peptides. Nevertheless, the simple synthesis of amino acids, the facile nature of their activation and condensation, their ability to recognize metals and cofactors and their remarkable capacity to self-assemble make peptides (and their analogues) favourable candidates for one of the earliest functional polymers. In this mini-review, we explore the ramifications of this hypothesis. Diverse lines of research in molecular biology, bioinformatics, geochemistry, biophysics and astrobiology provide clues about the progression and early evolution of proteins, and lend credence to the idea that early peptides served many central prebiotic roles before they were encodable by a polynucleotide template, in a putative ‘peptide-polynucleotide stage’. For example, early peptides and mini-proteins could have served as catalysts, compartments and structural hubs. In sum, we shed light on the role of early peptides and small proteins before and during the nucleotide world, in which nascent life fully grasped the potential of primordial proteins, and which has left an imprint on the idiosyncratic properties of extant proteins.