Exponentiated gradient algorithms for conditional random fields and max-margin Markov Networks

Log-linear and maximum-margin models are two commonly-used methods in supervised machine learning, and are frequently used in structured prediction problems. Efficient learning of parameters in these models is therefore an important problem, and becomes a key factor when learning from very large data sets. This paper describes exponentiated gradient (EG) algorithms for training such models, where EG updates are applied to the convex dual of either the log-linear or max-margin objective function; the dual in both the log-linear and max-margin cases corresponds to minimizing a convex function with simplex constraints. We study both batch and online variants of the algorithm, and provide rates of convergence for both cases. In the max-margin case, O(1/ε) EG updates are required to reach a given accuracy ε in the dual; in contrast, for log-linear models only O(log(1/ε)) updates are required. For both the max-margin and log-linear cases, our bounds suggest that the online EG algorithm requires a factor of n less computation to reach a desired accuracy than the batch EG algorithm, where n is the number of training examples. Our experiments confirm that the online algorithms are much faster than the batch algorithms in practice. We describe how the EG updates factor in a convenient way for structured prediction problems, allowing the algorithms to be efficiently applied to problems such as sequence learning or natural language parsing. We perform extensive evaluation of the algorithms, comparing them to L-BFGS and stochastic gradient descent for log-linear models, and to SVM-Struct for max-margin models. The algorithms are applied to a multi-class problem as well as to a more complex large-scale parsing task. In all these settings, the EG algorithms presented here outperform the other methods.

Mean and variance of estimates of the bispectrum of a harmonic random process-an analysis including leakage effects

Analytical expressions are derived for the mean and variance, of estimates of the bispectrum of a real-time series assuming a cosinusoidal model. The effects of spectral leakage, inherent in discrete Fourier transform operation when the modes present in the signal have a nonintegral number of wavelengths in the record, are included in the analysis. A single phase-coupled triad of modes can cause the bispectrum to have a nonzero mean value over the entire region of computation owing to leakage. The variance of bispectral estimates in the presence of leakage has contributions from individual modes and from triads of phase-coupled modes. Time-domain windowing reduces the leakage. The theoretical expressions for the mean and variance of bispectral estimates are derived in terms of a function dependent on an arbitrary symmetric time-domain window applied to the record. the number of data, and the statistics of the phase coupling among triads of modes. The theoretical results are verified by numerical simulations for simple test cases and applied to laboratory data to examine phase coupling in a hypothesis testing framework

Compilation of somatic mutations of the CDKN2 gene in human cancers : non-random distribution of base substitutions

The CDKN2 gene, encoding the cyclin-dependent kinase inhibitor p16, is a tumour suppressor gene that maps to chromosome band 9p21-p22. The most common mechanism of inactivation of this gene in human cancers is through homozygous deletion; however, in a smaller proportion of tumours and tumour cell lines intragenic mutations occur. In this study we have compiled a database of over 120 published point mutations in the CDKN2 gene from a wide variety of tumour types. A further 50 deletions, insertions, and splice mutations in CDKN2 have also been compiled. Furthermore, we have standardised the numbering of all mutations according to the full-length 156 amino acid form of p16. From this study we are able to define several hot spots, some of which occur at conserved residues within the ankyrin domains of p16. While many of the hotspots are shared by a number of cancers, the relative importance of each position varies, possibly reflecting the role of different carcinogens in the development of certain tumours. As reported previously, the mutational spectrum of CDKN2 in melanomas differs from that of internal malignancies and supports the involvement of UV in melanoma tumorigenesis. Notably, 52% of all substitutions in melanoma-derived samples occurred at just six nucleotide positions. Nonsense mutations comprise a comparatively high proportion of mutations present in the CDKN2 gene, and possible explanations for this are discussed.

Sequential release of BMP-7 and VEGF from the PLGA/AK-Gelatin Composite Scaffolds

Vascular endothelial growth factor (VEGF) and bone morphogenetic proteins (BMP-7) are key regulators of angiogenesis and osteogenesis during bone regeneration. The aim of this study was to investigate the possibility of realizing sequential release of the two growth factors using a novel composite scaffold. Poly(lactic-co-glycolic acid) (PLGA)-Akermanite (AK) microspheres were used to make the composite scaffold, which was then loaded with BMP-7, followed by embedding in a gelatin hydrogel matrix loaded with VEGF. The release profiles of the growth factors were studied and selected osteogenic related markers of bone marrow stromal cells (BMSCs) were analysed. It was shown that the composite scaffolds exhibited a fast initial burst release of VEGF within the first 3 days and a sustained slow release of BMP-7 over the full period of 20 days. The in vitro proliferation and differentiation of the BMSCs cultured in the osteogenic medium were enhanced by 1 to 2 times, resulting from the additionally and sequentially release of growth factors from the PLGA-AK/gelatin composite scaffolds.

Non-redundant sequential association rule mining based on closed sequential patterns

In many applications, e.g., bioinformatics, web access traces, system utilisation logs, etc., the data is naturally in the form of sequences. People have taken great interest in analysing the sequential data and finding the inherent characteristics or relationships within the data. Sequential association rule mining is one of the possible methods used to analyse this data. As conventional sequential association rule mining very often generates a huge number of association rules, of which many are redundant, it is desirable to find a solution to get rid of those unnecessary association rules. Because of the complexity and temporal ordered characteristics of sequential data, current research on sequential association rule mining is limited. Although several sequential association rule prediction models using either sequence constraints or temporal constraints have been proposed, none of them considered the redundancy problem in rule mining. The main contribution of this research is to propose a non-redundant association rule mining method based on closed frequent sequences and minimal sequential generators. We also give a definition for the non-redundant sequential rules, which are sequential rules with minimal antecedents but maximal consequents. A new algorithm called CSGM (closed sequential and generator mining) for generating closed sequences and minimal sequential generators is also introduced. A further experiment has been done to compare the performance of generating non-redundant sequential rules and full sequential rules, meanwhile, performance evaluation of our CSGM and other closed sequential pattern mining or generator mining algorithms has also been conducted. We also use generated non-redundant sequential rules for query expansion in order to improve recommendations for infrequently purchased products.

A sequential Monte Carlo approach to design for population pharmacokinetics studies

Here we present a sequential Monte Carlo approach that can be used to find optimal designs. Our focus is on the design of phase III clinical trials where the derivation of sampling windows is required, along with the optimal sampling schedule. The search is conducted via a particle filter which traverses a sequence of target distributions artificially constructed via an annealed utility. The algorithm derives a catalogue of highly efficient designs which, not only contain the optimal, but can also be used to derive sampling windows. We demonstrate our approach by designing a hypothetical phase III clinical trial.

Congo Red adsorption by ball-milled sugarcane bagasse

The adsorption of Congo Red (CR) by ball-milled sugarcane bagasse was evaluated in an aqueous batch system. CR adsorption capacity increased significantly with small changes in bagasse surface area. CR removal decreased with increasing solution pH from 5.0 to 10.0. Maximum adsorption capacity was 38.2 mg/g bagasse at a CR concentration of 500 mg/L. The equilibrium isotherm fitted the Freundlich model and the adsorption kinetics obeyed pseudo-second order equation. CR adsorption obeyed the intra-particle diffusion model very well with bagasse surface area in the range of 0.58–0.66 m2/g, whereas it was controlled by multi-adsorption stages with bagasse surface area in the range of 1.31–1.82 m2/g. Thermodynamic analysis indicated that the adsorption process is an exothermic and spontaneous process. Fourier transform infrared analysis of bagasse containing adsorbed CR indicated interactions between the carboxyl and hydroxyl groups of bagasse and CR function groups.

Influence of physical and chemical properties of solids on heavy metal adsorption

Partition of heavy metals between particulate and dissolve fraction of stormwater primarily depends on the adsorption characteristics of solids particles. Moreover, the bioavailability of heavy metals is also influenced by the adsorption behaviour of solids. However, due to the lack of fundamental knowledge in relation to the heavy metals adsorption processes of road deposited solids, the effectiveness of stormwater management strategies can be limited. The research study focused on the investigation of the physical and chemical parameters of solids on urban road surfaces and, more specifically, on heavy metal adsorption to solids. Due to the complex nature of heavy metal interaction with solids, a substantial database was generated through a series of field investigations and laboratory experiments. The study sites for the build-up pollutant sample collection were selected from four urbanised suburbs located in a major river catchment. Sixteen road sites were selected from these suburbs and represented typical industrial, commercial and residential land uses. Build-up pollutants were collected using a wet and dry vacuum collection technique which was specially designed to improve fine particle collection. Roadside soil samples were also collected from each suburb for comparison with the road surface solids. The collected build-up solids samples were separated into four particle size ranges and tested for a range of physical and chemical parameters. The solids build-up on road surfaces contained a high fraction (70%) of particles smaller than 150ìm, which are favourable for heavy metal adsorption. These solids particles predominantly consist of soil derived minerals which included quartz, albite, microcline, muscovite and chlorite. Additionally, a high percentage of amorphous content was also identified in road deposited solids. In comparing the mineralogical data of surrounding soil and road deposited solids, it was found that about 30% of the solids consisted of particles generated from traffic related activities on road surfaces. Significant difference in mineralogical composition was noted in different particle sizes of build-up solids. Fine solids particles (<150ìm) consisted of a clayey matrix and high amorphous content (in the region of 40%) while coarse particles (>150ìm) consisted of a sandy matrix at all study sites, with about 60% quartz content. Due to these differences in mineralogical components, particles larger than and smaller than 150ìm had significant differences in their specific surface area (SSA) and effective cation exchange capacity (ECEC). These parameters, in turn, exert a significant influence on heavy metal adsorption. Consequently, heavy metal content in >150ìm particles was lower than in the case of fine particles. The particle size range <75ìm had the highest heavy metal content, corresponding with its high clay forming minerals, high organic matter and low quartz content which increased the SSA, ECEC and the presence of Fe, Al and Mn oxides. The clay forming minerals, high organic matter and Fe, Al and Mn oxides create distinct groups of charge sites on solids surfaces and exhibit different adsorption mechanisms and bond strength, between heavy metal elements and charge sites. Therefore, the predominance of these factors in different particle sizes leads to different heavy metal adsorption characteristics. Heavy metals show preference for association with clay forming minerals in fine solids particles, whilst in coarse particles heavy metals preferentially associate with organic matter. Although heavy metal adsorption to amorphous material is very low, the heavy metals embedded in traffic related materials have a potential impact on stormwater quality.Adsorption of heavy metals is not confined to an individual type of charge site in solids, whereas specific heavy metal elements show preference for adsorption to several different types of charge sites in solids. This is attributed to the dearth of preferred binding sites and the inability to reach the preferred binding sites due to competition between different heavy metal species. This confirms that heavy metal adsorption is significantly influenced by the physical and chemical parameters of solids that lead to a heterogeneity of surface charge sites. The research study highlighted the importance of removal of solids particles from stormwater runoff before they enter into receiving waters to reduce the potential risk posed by the bioavailability of heavy metals. The bioavailability of heavy metals not only results from the easily mobile fraction bound to the solids particles, but can also occur as a result of the dissolution of other forms of bonds by chemical changes in stormwater or microbial activity. Due to the diversity in the composition of the different particle sizes of solids and the characteristics and amount of charge sites on the particle surfaces, investigations using bulk solids are not adequate to gain an understanding of the heavy metal adsorption processes of solids particles. Therefore, the investigation of different particle size ranges is recommended for enhancing stormwater quality management practices.

Effective adsorption of sodium dodecylsulfate (SDS) by hydrocalumite (CaAl-LDH-Cl) induced by self-dissolution and re-precipitation mechanism

Hydrocalumite (CaAl-LDH-Cl) were synthesized through a rehydration method involving a freshly prepared tricalcium aluminate (C3A) with CaCl2 solution. To understand the intercalation behaviour of sodium dodecylsulfate (SDS) with CaAl-LDH-Cl, X-ray diffraction (XRD), Fourier transform infrared (FTIR), scanning electron microscopy (SEM), transmission electron microscope (TEM), X-ray photoelectron spectroscopy (XPS), inductively coupled plasma-atomic emission spectrometer (ICP) and elemental analysis have been undertaken. The sorption isotherms with SDS reveal that the maximum sorption amount of SDS by CaAl-LDH-Cl could reach 3.67 mmol•g-1. The results revealed that CaAl-LDH-Cl holds a self-dissolution property, about 20-30% of which is dissolved. And the dissolved Ca2+, Al3+ ions are combined with SDS to form CaAl-SDS or Ca-SDS precipitation. It has been highlighted that the composition of resulting products is strongly dependent upon the SDS concentration. With increasing SDS concentrations, the main resulting product changes from CaAl-SDS to Ca-SDS, and the value of interlayer spacing increased to 3.27 nm.