Showing papers by "National University of Singapore published in 1997"

Feature Selection for Classification

Abstract: Feature selection has been the focus of interest for quite some time and much work has been done. With the creation of huge databases and the consequent requirements for good machine learning techniques, new problems arise and novel approaches to feature selection are in demand. This survey is a comprehensive overview of many existing methods from the 1970's to the present. It identifies four steps of a typical feature selection method, and categorizes the different existing methods in terms of generation procedures and evaluation functions, and reveals hitherto unattempted combinations of generation procedures and evaluation functions. Representative methods are chosen from each category for detailed explanation and discussion via example. Benchmark datasets with different characteristics are used for comparative study. The strengths and weaknesses of different methods are explained. Guidelines for applying feature selection methods are given based on data types and domain characteristics. This survey identifies the future research areas in feature selection, introduces newcomers to this field, and paves the way for practitioners who search for suitable methods for solving domain-specific real-world applications.

Comparing support vector machines with Gaussian kernels to radial basis function classifiers

Abstract: The support vector (SV) machine is a novel type of learning machine, based on statistical learning theory, which contains polynomial classifiers, neural networks, and radial basis function (RBF) networks as special cases. In the RBF case, the SV algorithm automatically determines centers, weights, and threshold that minimize an upper bound on the expected test error. The present study is devoted to an experimental comparison of these machines with a classical approach, where the centers are determined by X-means clustering, and the weights are computed using error backpropagation. We consider three machines, namely, a classical RBF machine, an SV machine with Gaussian kernel, and a hybrid system with the centers determined by the SV method and the weights trained by error backpropagation. Our results show that on the United States postal service database of handwritten digits, the SV machine achieves the highest recognition accuracy, followed by the hybrid system. The SV approach is thus not only theoretically well-founded but also superior in a practical application.

Empirics for Growth and Distribution: Stratification, Polarization, and Convergence Clubs

Abstract: This paper studies cross-country patterns of economic growth from the viewpoint of income distribution dynamics. Such a perspective raises new empirical and theoretical issues in growth analysis: the profound empirical regularity is an "emerging twin peaks" in the cross-sectional distribution, not simple patterns of convergence or divergence. The theoretical problems raised concern for interaction patterns among sub-groups of economies, not only problems of a single economy's accumulating factor inputs and technology growth.

Human DNA-(Cytosine-5) Methyltransferase-PCNA Complex as a Target for p21WAF1

Abstract: DNA–(cytosine-5) methyltransferase (MCMT) methylates newly replicated mammalian DNA, but the factors regulating this activity are unknown. Here, MCMT is shown to bind proliferating cell nuclear antigen (PCNA), an auxiliary factor for DNA replication and repair. Binding of PCNA requires amino acids 163 to 174 of MCMT, occurs in intact cells at foci of newly replicated DNA, and does not alter MCMT activity. A peptide derived from the cell cycle regulator p21WAF1 can disrupt the MCMT-PCNA interaction, which suggests that p21WAF1 may regulate methylation by blocking access of MCMT to PCNA. MCMT and p21WAF1 may be linked in a regulatory pathway, because the extents of their expression are inversely related in both SV40-transformed and nontransformed cells.

Liquid-Phase Microextraction in a Single Drop of Organic Solvent by Using a Conventional Microsyringe

Abstract: Two modes of liquid-phase microextraction (LPME) were developed for capillary gas chromatography. Both methodologies, i.e., static LPME and dynamic LPME, involve the use of very small amounts of organic solvent (<2 μL) in a conventional microsyringe. The performance of the two techniques is demonstrated in the determination of two chlorobenzenes extracted into a single drop of toluene by the use of a 10-μL syringe. Static LPME provided some enrichment (∼12-fold), good reproducibility (9.7%), and simplicity but suffered relatively long extraction time (15 min). Dynamic LPME provided higher (∼27-fold) enrichment within much shorter extraction time (∼3 min), and relatively poorer precision (12.8%), primarily due to repeated manual manipulation. Both methods allow the direct transfer of extracted analytes into a gas chromatograph.

Expression of constitutively active alpha-PAK reveals effects of the kinase on actin and focal complexes

Abstract: The family of p21-activated protein kinases (PAKs) appear to be present in all organisms that have Cdc42-like GTPases. In mammalian cells, PAKs have been implicated in the activation of mitogen-activated protein kinase cascades, but there are no reported effects of these kinases on the cytoskeleton. Recently we have shown that a Drosophila PAK is enriched in the leading edge of embryonic epithelial cells undergoing dorsal closure (N. Harden, J. Lee, H.-Y. Loh, Y.-M. Ong, I. Tan, T. Leung, E. Manser, and L. Lim, Mol. Cell. Biol. 16:1896-1908, 1996), where it colocalizes with structures resembling focal complexes. We show here by transfection that in epithelial HeLa cells alpha-PAK is recruited from the cytoplasm to distinct focal complexes by both Cdc42(G12V) and Rac1(G12V), which themselves colocalize to these sites. By deletion analysis, the N terminus of PAK is shown to contain targeting sequences for focal adhesions which indicate that these complexes are the site of kinase function in vivo. Cdc42 and Rac1 cause alpha-PAK autophosphorylation and kinase activation. Mapping alpha-PAK autophosphorylation sites has allowed generation of a constitutively active kinase mutant. By fusing regions of Cdc42 to the C terminus of PAK, activated chimeras were also obtained. Plasmids encoding these different constitutively active alpha-PAKs caused loss of stress fibers when introduced into both HeLa cells and fibroblasts, which was similar to the effect of introducing Cdc42(G12V) or Rac1(G12V). Significantly dramatic losses of focal adhesions were also observed. These combined effects resulted in retraction of the cell periphery after plasmid microinjection. These data support our previous suggestions of a role for PAK downstream of both Cdc42 and Rac1 and indicate that PAK functions include the dissolution of stress fibers and reorganization of focal complexes.

Long Polyglutamine Tracts in the Androgen Receptor Are Associated with Reduced Trans-Activation, Impaired Sperm Production, and Male Infertility

Abstract: The X-linked androgen receptor (AR) gene contains two polymorphic trinucleotide repeat segments that code for polyglutamine and polyglycine tracts in the N-terminal trans-activation domain of the AR protein. Changes in the lengths of these polymorphic repeat segments have been associated with increased risk of prostate cancer, an androgen-dependent tumor. Expansion of the polyglutamine tract causes a rare neuromuscular disease, spinal bulbar muscular atrophy, that is associated with low virilization, reduced sperm production, testicular atrophy, and infertility. As spermatogenesis is exquisitely androgen dependent, it is plausible that changes in these two repeat segments could have a role in some cases of male infertility associated with impaired spermatogenesis. To test this hypothesis, we examined the lengths of the polyglutamine and polyglycine repeats in 153 patients with defective sperm production and compared them to 72 normal controls of proven fertility. There was no significant association betwe...

Content-based retrieval of music and audio

Abstract: Though many systems exist for content-based retrieval of images, little work has been done on the audio portion of the multimedia stream. This paper presents a system to retrieve audio documents y acoustic similarity. The similarity measure is based on statistics derived from a supervised vector quantizer, rather than matching simple pitch or spectral characteristics. The system is thus able to learn distinguishing audio features while ignoring unimportant variation. Both theoretical and experimental results are presented, including quantitative measures of retrieval performance. Retrieval was tested on a corpus of simple sounds as well as a corpus of musical excerpts. The system is purely data-driven and does not depend on particular audio characteristics. Given a suitable parameterization, this method may thus be applicable to image retrieval as well.© (1997) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Manufacturer's Return Policies and Retail Competition

Abstract: Manufacturers' returns policies are a common feature in the distribution of many products. The obvious rationale for returns policies is insurance. Practitioners, not surprisingly, have a different perspective and view returns as a cost of doing business. In this paper, we study the strategic effect of returns policies on retail competition and highlight its profitability implications for a manufacturer. The setting for our research is the distribution of products with uncertain demand, limited shelf lives, and retail competition. Our objective is to provide a better understanding of when manufacturers should adopt returns policies. The insights are obtained with a model based on the economics of strategy and decision making under uncertainty. We show that when retailing is competitive and there is no uncertainty in demand, a returns policy subtly induces retailers to compete more intensely. The provision of a returns policy reduces retail prices without affecting wholesale prices, thereby reducing retailer margins and improving manufacturer profitability. The intuition is that with a returns policy, Cournot-like levels of retail stocks cannot be sustained. Each retailer will order stocks so that it will not be constrained by stocks, and so, intensifying retail competition. When, however, demand is uncertain and there is a single retailer, a returns policy encourages the retailer to over-stock, and so decreases upstream manufacturer profits. While the provision of a returns policy leads to lower retail margins, the optimality of returns policy depends on the overall uncertainty and marginal cost. A returns policy reduces the dispersion in retail prices between the high and low states of demand and the range of retail prices in the returns case is contained within the range of retail prices for the no-returns case. In the general setting, when there are competing retailers and demand is uncertain, there is a trade-off for the manufacturer between the benefits more intense retail competition and the costs excessive stocking of a returns policy. We find that a manufacturer should accept returns when the marginal production cost is sufficiently low and demand uncertainty is not too great. To validate the results of our theory, we conduct an empirical test with data from a major U.S. retailer. The tests confirm our prediction that a returns policy intensifies retail competition and reduces retailer margins. Our theory and empirical test should interest practitioners and researchers in distribution, especially those concerned with managing retail competition.

The core of an economy with multilateral environmental externalities

Abstract: When environmental externalities are international — i.e. transfrontier — they most often are multilateral and embody public good characteristics. Improving upon inefficient laissez-faire equilibria requires voluntary cooperation for which the game-theoretic core concept provides optimal outcomes that have interesting properties against free riding. To define the core, however, the characteristic function of the game associated with the economy (which specifies the payoff achievable by each possible coalition of players—here, the countries) must also reflect in each case the behavior of the players which are not members of the coalition. This has been for a long time a disputed issue in the theory of the core of economies with externalities.

Shape measures for content based image retrieval: a comparison

Abstract: A great deal of work has been done on the evaluation of information retrieval systems for alphanumeric data. The same thing can not be said about the newly emerging multimedia and image database systems. One of the central concerns in these systems is the automatic characterization of image content and retrieval of images based on similarity of image content. In this paper, we discuss effectiveness of several shape measures for content based similarity retrieval of images. The different shape measures we have implemented include outline based features (chain code based string features, Fourier descriptors, UNL Fourier features), region based features (invariant moments, Zernike moments, pseudo-Zernike moments), and combined features (invariant moments & Fourier descriptors, invariant moments & UNL Fourier features). Given an image, all these shape feature measures (vectors) are computed automatically, and the feature vector can either be used for the retrieval purpose or can be stored in the database for future queries. We have tested all of the above shape features for image retrieval on a database of 500 trademark images. The average retrieval efficiency values computed over a set of fifteen representative queries for all the methods is presented. The output of a sample shape similarity query using all the features is also shown.

Determination of biomass dry weight of marine microalgae

Abstract: Total biomass dry weight (DW) and ash free dry weight (AFDW) of five species of marine microalgae, Dunaliella sp., Isochrysis galbana, Nannochloropsis sp., Nitzschiaclosterium and Porphyridium cruentum, retained on filter paper, were determined. Dunaliella and Isochrysis cells have no cell wall; Nannochloropsis, Nitzschia and Porphyridium possess a cell wall and Nitzschia and Porphyridium cells are covered by silica and mucilage coating, respectively. In all these algae, DW of non-washed samples was at least 1.2 times higher than those washed by distilled water,0.9% sodium chloride, 0.5 M ammonium formate or 0.5 M ammonium bicarbonate. DW of 0.9% sodium chloride washed samples was more than 0.8 times higher than the other three washed samples. In most of the cases, there was no significant difference between DW of samples washed by ammonium formate and ammonium bicarbonate solutions (p>0.05). The AFDW of the non-washed algal samples was about twice that washed samples, and could be accounted for by volatile component in the sea water medium. Isotonic solution of ammonium bicarbonate is a satisfactory washing agent for algal cells for dry weight determination.

Neural-network feature selector

Abstract: Feature selection is an integral part of most learning algorithms. Due to the existence of irrelevant and redundant attributes, by selecting only the relevant attributes of the data, higher predictive accuracy can be expected from a machine learning method. In this paper, we propose the use of a three-layer feedforward neural network to select those input attributes that are most useful for discriminating classes in a given set of input patterns. A network pruning algorithm is the foundation of the proposed algorithm. By adding a penalty term to the error function of the network, redundant network connections can be distinguished from those relevant ones by their small weights when the network training process has been completed. A simple criterion to remove an attribute based on the accuracy rate of the network is developed. The network is retrained after removal of an attribute, and the selection process is repeated until no attribute meets the criterion for removal. Our experimental results suggest that the proposed method works very well on a wide variety of classification problems.

Towards an understanding of starch granule structure and hydrolysis

Abstract: Starch constitutes a major component of foods and also a raw material for use in the production of industrial products. The constituent macromolecules of starch are packaged in a highly ordered and compact manner, resulting in inert, insoluble granules. Food processing destroys starch structure, thereby releasing the component molecules, which are subsequently made available for hydrolysis or serve a functional role in the food system. The application of starch as a raw material usually requires the prior disruption of the inert granule structure, which involves additional processing steps. The hydrolysis of native granules has ramifications at all levels of the food processing chain, from postharvest losses to nutritional consequences of the ingested food. Technologists have not been able to take advantage of, or control, this process because the body of the information that is available on starch granule structure and the behaviour of hydrolysing enzymes is still incomplete. In this review, I have highlighted some of the more recent advances in this field, with the view to opening up the way for more efficient native starch hydrolysis.

Randomized Distributed Edge Coloring via an Extension of the Chernoff--Hoeffding Bounds

Abstract: Certain types of routing, scheduling, and resource-allocation problems in a distributed setting can be modeled as edge-coloring problems We present fast and simple randomized algorithms for edge coloring a graph in the synchronous distributed point-to-point model of computation Our algorithms compute an edge coloring of a graph $G$ with $n$ nodes and maximum degree $\Delta$ with at most $16 \Delta + O(\log^{1+ \delta} n)$ colors with high probability (arbitrarily close to 1) for any fixed $\delta > 0$; they run in polylogarithmic time The upper bound on the number of colors improves upon the $(2 \Delta - 1)$-coloring achievable by a simple reduction to vertex coloring To analyze the performance of our algorithms, we introduce new techniques for proving upper bounds on the tail probabilities of certain random variables The Chernoff--Hoeffding bounds are fundamental tools that are used very frequently in estimating tail probabilities However, they assume stochastic independence among certain random variables, which may not always hold Our results extend the Chernoff--Hoeffding bounds to certain types of random variables which are not stochastically independent We believe that these results are of independent interest and merit further study

Critical realism and realist research in human geography: a method or a philosophy in search of a method?

Abstract: Recent philosophical debates in human geography tend to misappropriate critical realism as a method per se. Drawing upon an extensive review of the realist philosophy and method in social science, this article argues that critical realism is a philosophy in search of a method. It first delves into recent debates about critical realism within the wider geographical discourse. It then suggests three useful guidelines in executing realist research in human geography: iterative abstraction, qualified grounded theory method and methodological triangulation. The article ends with a detailed empirical example for the readers to work through of the ways in which realist research can be practised in human geography.

Frequency of formation of chimeric molecules as a consequence of PCR coamplification of 16S rRNA genes from mixed bacterial genomes.

Abstract: PCR is routinely used in amplification and cloning of rRNA genes from environmental DNA samples for studies of microbial community structure and identification of novel organisms. There have been concerns about generation of chimeric sequences as a consequence of PCR coamplification of highly conserved genes, because such sequences may lead to reports of nonexistent organisms. To quantify the frequency of chimeric molecule formation, mixed genomic DNAs from eight actinomycete species whose 16S rRNA sequences had been determined were used for PCR coamplification of 16S rRNA genes. A large number of cloned 16S ribosomal DNAs were examined by sequence analysis, and chimeric molecules were identified by multiple-sequence alignment with reference species. Here, we report that the level of occurrence of chimeric sequences after 30 cycles of PCR amplification was 32%. We also show that PCR-induced chimeras were formed between different rRNA gene copies from the same organism. Because of the wide use of PCR for direct isolation of 16S rRNA sequences from environmental DNA to assess microbial diversity, the extent of chimeric molecule formation deserves serious attention.

Feature selection via discretization

Abstract: Discretization can turn numeric attributes into discrete ones. Feature selection can eliminate some irrelevant and/or redundant attributes. Chi2 is a simple and general algorithm that uses the /spl chi//sup 2/ statistic to discretize numeric attributes repeatedly until some inconsistencies are found in the data. It achieves feature selection via discretization. It can handle mixed attributes, work with multiclass data, and remove irrelevant and redundant attributes.

Optimising train movements through coast control using genetic algorithms

Abstract: A genetic algorithm (GA) is proposed to optimise train movements using appropriate coast control that can be integrated within automatic train operation (ATO) systems. The coast control output for a train changes with the interstation distances and gradient profiles, and the current operating conditions of the mass rapid transit (MRT) system, namely: (i) train schedules; (ii) expected passenger loads; and (iii) expected track voltages. The algorithm generates an optimum coast control based on evaluation of the punctuality, riding comfort and energy consumption. Before the train sets off to the designated station, a coast control table is generated that will be referenced by the train at runtime for deciding when to initiate coasting or resume motoring control. Each coast control table is encoded into variable length chromosomes with each gene representing the relative position between stations where coasting should be initiated or terminated. Each generation is evolved from mating of the paired equal-length chromosomes with possibilities of crossover, mutations, gene duplications and gene deletions. The key feature of this method is that it has a solid mathematical foundation. Effectively, the implementation provides good, credible and reasonably fast solutions for this variable dimensional and multiobjective optimisation problem. The algorithm has the potential for online implementation for producing a coast control lookup table for each interstation run before the train sets off. The results, although preliminary, suggest that the method is promising.

Synthesis, Characterization, and Nonlinear Optical Properties of Copper Nanoparticles

Abstract: In this paper, copper nanoparticles were prepared by the reduction of copper(II) acetate in water and 2-ethoxyethanol using hydrazine under reflux. The synthesized nanoparticles exhibit a distinct absorption peak in the region 572−582 nm. The average size variation from 6.6 to 22.7 nm in ethoxyethanol and from 15.5 to 30.2 nm in water was achieved by the addition of various amounts of a protective polymer (poly(N-vinylpyrrolidone)). The nonlinear optical properties of the copper colloids were first measured using the Z-scan technique. The χ(3)/α0 values obtained were found to be of the magnitude of 10-11−10-12 esu cm, which are in good agreement with the reported values obtained for copper nanoparticles embedded in glass.

The Impact of Technological Complexity and Interfirm Cooperation on Business Survival

Abstract: This study's predictions are that businesses developing high-complexity technologies face higher risks of failure than other businesses because of greater competency demands and higher organization...

Soft handoffs in CDMA mobile systems

Abstract: This article presents an overview of soft handoff, an idea which is becoming quite important because of its use in the IS-95 code-division multiple access (CDMA) cellular phone standard. The benefits and disadvantages of using soft handoff over hard handoff are discussed, with most results drawn from the available literature. The two most well-known benefits are fade margin improvement and higher uplink capacity, while disadvantages include increased downlink interference and more complex implementation. Handoff parameter optimization is extremely important, so various studies on the trade-offs to be considered when selecting these parameters are surveyed, from both the link quality and resource allocation perspectives. Finally, research directions and future trends are discussed.