Part and Attribute Discovery from Relative Annotations

Part and attribute based representations are widely used to support high-level search and retrieval applications. However, learning computer vision models for automatically extracting these from images requires significant effort in the form of part and attribute labels and annotations. We propose an annotation framework based on comparisons between pairs of instances within a set, which aims to reduce the overhead in manually specifying the set of part and attribute labels. Our comparisons are based on intuitive properties such as correspondences and differences, which are applicable to a wide range of categories. Moreover, they require few category specific instructions and lead to simple annotation interfaces compared to traditional approaches. On a number of visual categories we show that our framework can use noisy annotations collected via “crowdsourcing” to discover semantic parts useful for detection and parsing, as well as attributes suitable for fine-grained recognition.     

Spinning of poly(ethylene terephthalate) fibers from a melt pool without a spinneret

Monofilament fibers were spun continuously from the free surface of a pool of molten poly(ethylene terephthalate) without the aid of a spinneret. For take-up velocities in the range of 12 to 400 ft/min, the denier of the filaments produced was an inverse power function of take-up velocity, and the birefringence was an inverse power function of the filament diameter. Production rate and product uniformity were strongly dependent on take-up velocity and surface temperature of the melt pool.     

A MECHANISTICALLY BASED MATHEMATICAL MODEL OF SULFUR DIOXIDE ABSORPTION INTO A CALCIUM HYDROXIDE SLURRY IN A SPRAY DRYER

A mathematical model has been developed to predict So₂ absorption and removal during the constant rate drying period of a spray dryer. The model, based on film theory, treats the atomized slurry droplets as spheres containing discrete sorbent particles of slaked lime with the fluid uniformly distributed around the individual particles. The model includes gas and liquid phase mass transfer coefficients as well as resistance to Ca(OH)₂ dissolution. A sensitivity analysis has been conducted and a comparison was made between pilot-scale experimental data and model-predicted values of S0₂ removal efficiency.     

Substrate-mediated interactions and intermolecular forces between molecules adsorbed on surfaces

Adsorbate interactions and reactions on metal surfaces have been investigated using scanning tunneling microscopy. The manners in which adsorbates perturb the surface electronic structure in their vicinity are discussed. The effects these perturbations have on other molecules are shown to be important in overlayer growth. Interactions of molecules with surface steps are addressed, and each molecule's electron affinity is shown to dictate its adsorption sites at step edges. Standing waves emanating from steps are demonstrated to effect transient molecular adsorption up to 40 A away from the step edge. Halobenzene derivatives are used to demonstrate how the surface is important in aligning reactive intermediates.     

Relationship of peroxide value and thiobarbituric acid value to development of undesirable flavor characteristics in fats

The peroxide value and thiobarbituric acid value were compared with the flavor score for a series of different types of fats, with and without added monoglyceride and with and without different stabilizers. The data indicated that the flavor score cannot be estimated for any given fat from either the peroxide value or the thiobarbituric acid value. Either can be used to follow the development of off flavors in a given product or formulation but the relative level may vary from product to product. Presented at the AOCS Meeting in Minneapolis, 1963.     

Quantum correlations at negative absolute temperatures

We study behavior of quantum discord, a kind of quantum correlation, in systems of dipole–dipole interacting spins in an external magnetic field in the whole temperature range ( \(-\infty ). It was shown that negative temperatures, which are introduced to describe inversions in the population in a finite level system, provide more favorable conditions for emergence of quantum correlations including entanglement. We show that at negative temperature, the correlations become more intense and discord exists between remote spins being in separated states.     

Biodesulfurization of dibenzothiophene sulfone by Arthrobacter sp. and studies with oxidized Illinois No. 6 coal

A bacterium identified as Arthrobacter sp. was grown on dibenzothiophene (DBT) sulfone as a sole source of sulfur, producing sulfite and sulfate. Sulfur in DBT sulfone (1.0 mM) was nearly quantitatively converted to sulfate by the organism. The organism could also use DBT sulfone as a sole source of carbon and energy. There was evidence for transient accumulation of benzoic acid in the culture medium after growth of the cells slowed. The DBT sulfoxide analogue 9-fluorenone was converted by resting cells to a product identified as 1,10-dihydroxy-1,10-dihydrofluoren-9-one, suggesting that DBT sulfone may be metabolized via an angular hydroxylation resulting in carbon- sulfur bond cleavage. This strain of Arthrobacter showed no ability to desulfurize oxidized Illinois No. 6 coal.     

Creep of 6H α-Silicon Carbide Single Crystals

The compressive creep behavior of single-crystal 6H α-SiC was measured for orientations parallel to and at 45° to [0001]. Deformation of the 45° orientation was dominated by basal slip. Steady-state creep rates above 10^-7/s were measured at temperatures as low as 800°C. An activation energy of 277 kJ/mol and a stress exponent of 3.32 were determined. Creep testing with applied stresses parallel to [0001] was performed at 1650°C to 1850°C, yielding a stress exponent and activation energy of 4.93 and 180 kJ/mol, respectively. The occurrence of basal slip in the [0001] specimens suggested that significant off-axis stresses were present during testing.     

Rheological properties of molten poly(ethylene terephthalate)

The complete steady-state flow properties of molten poly(ethylene terephthalate) for shear stresses ≦4.14 × 10⁶ dynes/cm² were determined. A single, complete master curve had been constructed in earlier work by Gregory and Watson; the curve interrelates the shear stress, shear rate, temperature, and molecular weight (inherent viscosity) by using a temperature superposition scheme from the literature and a similar molecular weight superposition scheme. Equations in agreement with theory and with other published experimental data were derived from the master curve. Results presented here make possible the direct calculation of the melt viscosity of poly(ethylene terephthalate) at shear stresses ≦4.14 × 10⁶ dynes/cm². The effects of a unit temperature change and/or a unit change in inherent viscosity (I. V.) on the melt viscosity were determined. For poly(ethylene terephthalate) with a 0.6 I. V., a 0.0025 change in I. V. accounts for about the same change in melt viscosity as a 1°C change in temperature.     

An ontology-driven framework towards building enterprise semantic information layer

Enterprises Information Systems (EIS) have been applied for decades in Computer-Aided Engineering (CAE) and Computer-Aided Design (CAD), where huge amount of increasing data is stored in the heterogeneous and distributed systems. As systems evaluating, system redesign and reengineering are demanded. A facing challenge is how to interoperate among different systems by overcoming the gap of conceptual heterogeneity. In this article, an enlarged data representation called semantic information layer (SIL) is described for facilitating heterogeneous systems interoperable. SIL plays a role as mediation media and knowledge representation among heterogeneous systems. The SIL building process is based on ontology engineering, including ontology extraction from relational database (RDB), ontology enrichment and ontology alignment. Mapping path will maintain the links between SIL and data source, and query implementation and user interface are for retrieving data and interacting with end users. We described fully a practical ontology-driven framework for building SIL and introduced extensively relevant standards and techniques for implementing the framework. In the core part of ontology development, a dynamic multi-strategies ontology alignment with automatic matcher selection and dynamic similarity aggregation is proposed. A demonstration case study in the scenario of mobile phone industry is used to illustrate the proposed framework.