Distance metric learning for graph structured data

Machine Learning - Graphs are versatile tools for representing structured data. As a result, a variety of machine learning methods have been studied for graph data analysis. Although many such...     

Molecular aggregation of PEEK with PES blends and the block copolymers composed of PEEK and PES components

Polymer blends of PEEK with PES were prepared by the solution blending method. Copolymers composed of PEEK and PES components were synthesized from these oligomers. The formation conditions exerted an influence over the molecular aggregation and the crystallization behaviors of the blend films and block copolymers which were examined by X-ray diffractometry and DSC analysis. As a result, phase-separation in the blend films was found when the formation temperature was high. The blend films formed at 340°C, quenched and annealed at 180°C, exhibited the same crystallization behavior as those of PEEK. In the case of the blend films formed at 300°C, the annealing of the films at 250°C was required to crystallize the blend films. The T_g of a copolymer with a PEEK component content of more than 50% tends to shift toward a higher temperature than the T_g of PEEK itself, and the T_m of the copolymer toward a lower temperature than that of PEEK ist.     

Preparation of porous carbons from thermoplastic precursors and their performance for electric double layer capacitors

Porous carbons with high surface area were successfully prepared from thermoplastic precursors, such as poly(vinyl alcohol) (PVA), hydroxyl propyl cellulose and poly(ethylene terephthalate), by the carbonization of a mixture with MgO at 900 °C in an inert atmosphere. After carbonization the MgO was dissolved out using a diluted sulfuric acid and the carbons formed were isolated. The mixing of the PVA carbon precursor with the MgO precursors (reagent grade MgO, magnesium acetate or citrate) was done either in powder form or in an aqueous solution. The BET surface area of the carbons obtained via solution mixing could reach a very high value, such as 2000 m²/g, without any activation process. The pore structure of the resultant carbons was found to depend strongly on the mixing method; the carbons prepared via solution mixing were rich in mesopores, but those produced via powder mixing were rich in micropores. The size of mesopores was found to be almost the same as that of the MgO particles, suggesting a way of controlling the mesopore size in the resultant carbons. Measurement of capacitance was carried out in 1 mol/L H₂SO₄ electrolyte. The porous carbon with a BET surface area of 1900 m²/g prepared at 900 °C through solution mixing of Mg acetate with PVA showed a fairly high EDLC capacitance, about 250 F/g with a current density of 20 mA/g and 210 F/g with 1000 mA/g. The rate performance was closely related to the mesoporous surface area.     

Hydrothermal Formation of Hydroxyapatite Layers on the Surface of Type-A Zeolite

Type-A zeolite evenly covered with hydroxyapatite thin layers was prepared using hydrothermal treatment at 120°C for 8 h under autogenous pressure. The hydroxyapatite needlelike nanocrystals, 100–200 nm in diameter and 30 nm in thickness, were grown under the reaction between discharged Ca²⁺ ions from type-A zeolite and PO₄³⁻ ions in (NH₄)₃PO₄ solution. The preferential orientations of the c -axis of hydroxyapatite crystals perpendicular to a zeolite surface were observed using transmission electron microscopy. The crystal structure of type-A zeolite was not destroyed under the reaction, but the surface morphology was changed only with complete covering of scaly hydroxyapatite particles.     

Preparation of gelatinous aluminium hydroxide by urea from aqueous solutions containing chloride,nitrate and sulphate of aluminium

The preparation of gelatinous aluminium hydroxide from aqueous solutions containing a mixture of the chloride, nitrate or sulphate of aluminium and urea by heating at 95°C has been investigated under different conditions. The pH value of aqueous solutions, on heating for a given period of time, gradually increases, rises steeply at pH 4.0–7.5 and finally approaches a constant value. The precipitate appears at about pH 7 in the presence of chloride or nitrate ions and about pH 4 in the presence of sulphate ions. Although the gelatinous precipitates in the chloride and nitrate systems are apparently different from the granular, filterable one in the sulphate system, their compositions are not influenced by the species of aluminium salt. The fresh precipitates exist in an amorphous state, and go to pseudoboehmite by ageing. It seems that the amount of pseudoboehmite increases as the concentration of aluminium salt in aqueous solution decreases. However, the transformation from amorphous aluminium hydroxide to pseudoboehmite is reduced in the presence of sulphate ions. Furthermore, it is found that the X-ray diffraction peak for the (200) plane, as an orthorhombic structure, in the pseudoboehmite precipitated from chloride or nitrate solution is more intense than that from sulphate solution.     

500-fps face tracking system

In this paper, we propose a high-speed vision system that can be applied to real-time face tracking at 500 fps using GPU acceleration of a boosting-based face tracking algorithm. By assuming a small image displacement between frames, which is a property of high-frame rate vision, we develop an improved boosting-based face tracking algorithm for fast face tracking by enhancing the Viola–Jones face detector. In the improved algorithm, face detection can be efficiently accelerated by reducing the number of window searches for Haar-like features, and the tracked face pattern can be localized pixel-wise even when the window is sparsely scanned for a larger face pattern by introducing skin color extraction in the boosting-based face detector. The improved boosting-based face tracking algorithm is implemented on a GPU-based high-speed vision platform, and face tracking can be executed in real time at 500 fps for an 8-bit color image of 512 × 512 pixels. In order to verify the effectiveness of the developed face tracking system, we install it on a two-axis mechanical active vision system and perform several experiments for tracking face patterns.