Dynamic behavior of balanced NV-trees

In recent years, some approximate high-dimensional indexing techniques have shown promising results by trading off quality guarantees for improved query performance. While the query performance and quality of these methods has been well studied, however, the performance of index maintenance has not yet been reported in any detail. Here, we focus on the dynamic behavior of the balanced NV-tree, which is a disk-based approximate index for very large collections. We report on an initial study of the effects of several implementation choices for the balanced NV-tree, and show that with appropriate implementation, significant performance improvements are possible. Overall, the proposed techniques not only reduce maintenance cost, but can also improve search performance significantly with minimal loss of search quality.     

Room temperature deposition of self-assembled Al nanoclusters on stepped sapphire (0001) surface and subsequent nitridation

Self-assembled growth and nitridation of ultrathin Al nanoclusters on a stepped sapphire (0001) surface were studied by high-resolution X-ray photoemission spectroscopy, atomic force microscopy and low-energy electron diffraction (LEED). Upon room temperature deposition, in the coverage range of ∼ 0.79 to 2.3 monolayer (ML), Al nanoclusters were uniformly nucleated over the entire surface of defect-free atomically smooth terraces as well as step edges. Subsequent nitridation at elevated temperatures by ammonia did not alter the morphology of the nanoclusters. The global morphology of the stepped sapphire (0001) surface such as terrace width, step height and facet orientation had no obvious influence on the nucleation morphology of the nanoclusters in the given Al coverage range. However, local structural defects at the joints of short facets and step edges played a noticeable role on the local morphology of the nanoclusters and subsequently the nitridation chemistry. The Al nanoclusters were uniformly nitridated from surface and downwards through the 3D structures. The LEED pattern indicated a certain degree of crystallinity on the nitridated surface at a nominal Al coverage less than 2 ML, whereas at 2.3 ML Al coverage, the nitridated surface became amorphous. Thus there is a critical coverage for good surface order.     

Optimization-based feature selection with adaptive instance sampling

Preprocessing the data to filter out redundant and irrelevant features is one of the most important steps in the data mining process. Careful feature selection may improve both the computational time of inducing subsequent models and the quality of those models. Using fewer features often leads to simpler and easier to interpret models, and selecting important feature can lead to important insights into the application. The feature selection problem is inherently a combinatorial optimization problem. This paper builds on a metaheuristic called the nested partitions method that has been shown to be particularly effective for the feature selection problem. Specifically, we focus on the scalability of the method and show that its performance is vastly improved by incorporating random sampling of instances. Furthermore, we develop an adaptive variant of the algorithm that dynamically determines the required sample rate. The adaptive algorithm is shown to perform very well when applied to a set of standard test problems.     

Classification for predicting offender affiliation with murder victims

The National Incident-Based Reporting System (NIBRS) is used by law enforcement to record a detailed picture of crime incidents, including data on offenses, victims and suspected arrestees. Such incident data lends itself to the use of data mining to uncover hidden patterns that can provide meaningful insights to law enforcement and policy makers. In this paper we analyze all homicide data recorded over one year in the NIBRS database, and use classification to predict the relationships between murder victims and the offenders. We evaluate different ways for formulating classification problems for this prediction and compare four classification methods: decision tree, random forest, support vector machine and neural network. Our results show that by setting up binary classification problems to discriminate each type of victim–offender relationship versus all others good classification accuracy can be obtained, especially by the support vector machine method and the random forest approach. Furthermore, our results show that interesting structural insight can be obtain by performing attribute selection and by using transparent decision tree models.     

Detection of skin cancer by classification of Raman spectra

Skin lesion classification based on in vitro Raman spectroscopy is approached using a nonlinear neural network classifier. The classification framework is probabilistic and highly automated. The framework includes a feature extraction for Raman spectra and a fully adaptive and robust feedforward neural network classifier. Moreover, classification rules learned by the neural network may be extracted and evaluated for reproducibility, making it possible to explain the class assignment. The classification performance for the present data set, involving 222 cases and five lesion types, was 80.5%+/-5.3% correct classification of malignant melanoma, which is similar to that of trained dermatologists based on visual inspection. The skin cancer basal cell carcinoma has a classification rate of 95.8%+/-2.7%, which is excellent. The overall classification rate of skin lesions is 94.8%+/-3.0%. Spectral regions, which are important for network classification, are demonstrated to reproduce. Small distinctive bands in the spectrum, corresponding to specific lipids and proteins, are shown to hold the discriminating information which the network uses to diagnose skin lesions.     

Discovering Dispatching Rules Using Data Mining

This paper introduces a novel methodology for generating scheduling rules using a data-driven approach. We show how to use data mining to discover previously unknown dispatching rules by applying the learning algorithms directly to production data. This approach involves preprocessing of historic scheduling data into an appropriate data file, discovery of key scheduling concepts, and representation of the data mining results in a way that enables its use for job scheduling. We also consider how by using this new approach unexpected knowledge and insights can be obtained, in a manner that would not be possible if an explicit model of the system or the basic scheduling rules had to be obtained beforehand. All of our results are illustrated via numerical examples and experiments on simulated data.     

Vertical Phase Separation in Small Molecule:Polymer Blend Organic Thin Film Transistors Can Be Dynamically Controlled

Blending of small‐molecule organic semiconductors (OSCs) with amorphous polymers is known to yield high performance organic thin film transistors (OTFTs). Vertical stratification of the OSC and polymer binder into well‐defined layers is crucial in such systems and their vertical order determines whether the coating is compatible with a top and/or a bottom gate OTFT configuration. Here, we investigate the formation of blends prepared via spin‐coating in conditions which yield bilayer and trilayer stratifications. We use a combination of in situ experimental and computational tools to study the competing effects of formulation thermodynamics and process kinetics in mediating the final vertical stratification. It is shown that trilayer stratification (OSC/polymer/OSC) is the thermodynamically favored configuration and that formation of the buried OSC layer can be kinetically inhibited in certain conditions of spin‐coating, resulting in a bilayer stack instead. The analysis reveals here that preferential loss of the OSC, combined with early aggregation of the polymer phase due to rapid drying, inhibit the formation of the buried OSC layer. The fluid dynamics and drying kinetics are then moderated during spin‐coating to promote trilayer stratification with a high quality buried OSC layer which yields unusually high mobility >2 cm² V^?1 s^?1 in the bottom‐gate top‐contact configuration.     

Chemical composition of arctic snow: concentration levels and regional distribution of major elements

At the end of the northern winter 1996/1997, 21 snow samples were collected from 17 arctic localities in Norway, Sweden, Finland, Svalbard, Russia, Alaska, Canada, Greenland and Iceland. Major element concentrations of the filtered (0.45 mum) melted snow indicate that most samples are consistent with a diluted seawater composition. Deviations from this behaviour indicate additional SO(4)(2-) and Cl(-) relative to seawater, suggesting a minor contribution from (probably local) coal combustion emissions (Alaska, Finland, Sweden, Svalbard). The samples with the highest Na and Cl(-) content (Canada, Russia) also have higher Na/SO(4)(2-) and Cl(-)/SO(4)(2-) ratios than seawater, suggesting a slight contamination from (probably local) deicing activities. Local soil or rock dust inputs in the snow are indicated by 'excess' Ca contents (Alaska, Svalbard, Greenland, Sweden). No overall relationship was found between pH (range: 4.6-6.1) and total or non-seasalt SO(4)(2-) (NSS), suggesting that acidification due to long-range transport of SO(2) pollution is not operating on an arctic-wide scale. In a few samples (Alaska, Finland, Sweden, Svalbard), a significant proportion (>50%) of SO(4)(2-) is non-marine in origin. Sources for this non-marine SO(4)(2-) need not all be found in long-range atmospheric transport and more likely sources are local industry (Finland, Sweden), road traffic (Alaska) or minor snow-scooting traffic (one Svalbard locality). A few samples from northern Europe show a relatively weak trend of decreasing pH with increasing NO(3)(-).     

The impact of sampling techniques on soil pore water carbon measurements of an Icelandic Histic Andosol

The carbon in soil pore water from a Histic Andosol from Western Iceland was studied at three different scales; in the field, in undisturbed outdoor mesocosms and in laboratory repacked microcosms. Pore water was extracted using suction cup lysimeters and hollow-fibre tube sampler devices (Rhizon samplers). There were significant differences in all measured variables, dissolved inorganic carbon (DIC), dissolved organic carbon (DOC) and pH values between the scales of the experiment. Gaseous constituents of soil solution and pH were more susceptible to changes in scale and the type of sampling devices used. Dissolved inorganic carbon concentrations did not differ significantly between field and mesocosm solutions but where up to 14 times lower in microcosms compared to mesocosms solutions. Rhizon samplers yielded solutions with up to 4.7 times higher DIC concentrations than porous cup lysimeters. Mesocosm surface horizon DOC concentrations were 20 and 2 times higher than in field and microcosms respectively. There was difference in DOC concentration between sampling methods (up to 8 times higher in suction cups than rhizon samplers) above 50 cm depth. Soil solution pH values did not differ between field and mesocosms and mesocosms and microcosms respectively down to 80 cm depth. Direct comparison between field and microcosms was not possible due to the nature of sampling devices. Soil solutions sampled with Rhizon samplers yielded lower pH values (up to 1.3 pH units) than those sampled with suction cups. Twenty percent of annually bound organic carbon at the soils surface under field conditions was lost by leaching of DOC and through decomposition to DIC in disturbed non-vegetated microcosms. This percentage increased to 38% in undisturbed vegetated mesocosms highlighting the importance of surface vegetation in importing carbon to soils. Increased influx of nutrients will increase growth and photosynthesis but decrease carbon sequestration in near surface horizons. Although field studies considering long-term anthropogenic changes in pedogenesis require considerable experimental duration, more rapid experiments can be conducted with confidence in micro- and mesocosms as in this research.