Design of multiple sequence alignment algorithms on parallel, distributed memory supercomputers

The challenge of comparing two or more genomes that have undergone recombination and substantial amounts of segmental loss and gain has recently been addressed for small numbers of genomes. However, datasets of hundreds of genomes are now common and their sizes will only increase in the future. Multiple sequence alignment of hundreds of genomes remains an intractable problem due to quadratic increases in compute time and memory footprint. To date, most alignment algorithms are designed for commodity clusters without parallelism. Hence, we propose the design of a multiple sequence alignment algorithm on massively parallel, distributed memory supercomputers to enable research into comparative genomics on large data sets. Following the methodology of the sequential progressiveMauve algorithm, we design data structures including sequences and sorted k-mer lists on the IBM Blue Gene/P supercomputer (BG/P). Preliminary results show that we can reduce the memory footprint so that we can potentially align over 250 bacterial genomes on a single BG/P compute node. We verify our results on a dataset of E.coli, Shigella and S.pneumoniae genomes. Our implementation returns results matching those of the original algorithm but in 1/2 the time and with 1/4 the memory footprint for scaffold building. In this study, we have laid the basis for multiple sequence alignment of large-scale datasets on a massively parallel, distributed memory supercomputer, thus enabling comparison of hundreds instead of a few genome sequences within reasonable time.

Learning Bayesian networks in Semi-deterministic systems

In current constraint-based (Pearl-style) systems for discovering Bayesian networks, inputs with deterministic relations are prohibited. This restricts the applicability of these systems. In this paper, we formalize a sufficient condition under which Bayesian networks can be recovered even with deterministic relations. The sufficient condition leads to an improvement to Pearl’s IC algorithm; other constraint-based algorithms can be similarly improved. The new algorithm, assuming the sufficient condition proposed, is able to recover Bayesian networks with deterministic relations, and moreover suffers no loss of performance when applied to nondeterministic Bayesian networks.

Real-time lane detection on suburban streets using visual cue integration

The detection of lane boundaries on suburban streets using images obtained from video constitutes a challenging task. This is mainly due to the difficulties associated with estimating the complex geometric structure of lane boundaries, the quality of lane markings as a result of wear, occlusions by traffic, and shadows caused by road-side trees and structures. Most of the existing techniques for lane boundary detection employ a single visual cue and will only work under certain conditions and where there are clear lane markings. Also, better results are achieved when there are no other onroad objects present. This paper extends our previous work and discusses a novel lane boundary detection algorithm specifically addressing the abovementioned issues through the integration of two visual cues. The first visual cue is based on stripe-like features found on lane lines extracted using a two-dimensional symmetric Gabor filter. The second visual cue is based on a texture characteristic determined using the entropy measure of the predefined neighbourhood around a lane boundary line. The visual cues are then integrated using a rulebased classifier which incorporates a modified sequential covering algorithm to improve robustness. To separate lane boundary lines from other similar features, a road mask is generated using road chromaticity values estimated from CIE L*a*b* colour transformation. Extraneous points around lane boundary lines are then removed by an outlier removal procedure based on studentized residuals. The lane boundary lines are then modelled with Bezier spline curves. To validate the algorithm, extensive experimental evaluation was carried out on suburban streets and the results are presented. 

An improved Gbest guided artificial bee colony (IGGABC) algorithm for classification and prediction tasks

Artificial Neural Networks (ANN) performance depends on network topology, activation function, behaviors of data, suitable synapse's values and learning algorithms. Many existing works used different learning algorithms to train ANN for getting high performance. Artificial Bee Colony (ABC) algorithm is one of the latest successfully Swarm Intelligence based technique for training Multilayer Perceptron (MLP). Normally Gbest Guided Artificial Bee Colony (GGABC) algorithm has strong exploitation process for solving mathematical problems, however the poor exploration creates problems like slow convergence and trapping in local minima. In this paper, the Improved Gbest Guided Artificial Bee Colony (IGGABC) algorithm is proposed for finding global optima. The proposed IGGABC algorithm has strong exploitation and exploration processes. The experimental results show that IGGABC algorithm performs better than that standard GGABC, BP and ABC algorithms for Boolean data classification and time-series prediction tasks.

An improved soft-kill BESO algorithm for optimal distribution of single or multiple material phases

Finding the optimum distribution of material phases in a multi-material structure is a frequent and important problem in structural engineering which involves topology optimization. The Bi-directional Evolutionary Structural Optimization (BESO) method is now a well-known topology optimization method. In this paper an improved soft-kill BESO algorithm is introduced which can handle both single and multiple material distribution problems. A new filtering scheme and a gradual procedure inspired by the continuation approach are used in this algorithm. Capabilities of the proposed method are demonstrated using different examples. It is shown that the proposed method can result in considerable improvements compared to the normal BESO algorithm particularly when solving problems involving very soft material or void phase.

IncSPADE: an incremental sequential pattern mining algorithm based on SPADE property

In this paper we propose Incremental Sequential PAttern Discovery using Equivalence classes (IncSPADE) algorithm to mine the dynamic database without the requirement of re-scanning the database again. In order to evaluate this algorithm, we conducted the experiments against three different artificial datasets. The result shows that IncSPADE outperformed the benchmarked algorithm called SPADE up to 20%.

Fast algorithm for the cutting angle method of global optimization

The cutting angle method for global optimization was proposed in 1999 by Andramonov et al. (Appl. Math. Lett. 12 (1999) 95). Computer implementation of the resulting algorithm indicates that running time could be improved with appropriate modifications to the underlying mathematical description. In this article, we describe the initial algorithm and introduce a new one which we prove is significantly faster at each stage. Results of numerical experiments performed on a Pentium III 750 Mhz processor are presented.

Scheduling of a parallel computation-bound application and sequential applications executing concurrently on a cluster - a case study

Studies have shown that most of the computers in a non-dedicated cluster are often idle or lightly loaded. The underutilized computers in a non-dedicated cluster can be employed to execute parallel applications. The aim of this study is to learn how concurrent execution of a computation-bound and sequential applications influence their execution performance and cluster utilization. The result of the study has demonstrated that a computation-bound parallel application benefits from load balancing, and at the same time sequential applications suffer only an insignificant slowdown of execution. Overall, the utilization of a non-dedicated cluster is improved.

Yet another induction algorithm

Inducing general functions from specific training examples is a central problem in the machine learning. Using sets of If-then rules is the most expressive and readable manner. To find If-then rules, many induction algorithms such as ID3, AQ, CN2 and their variants, were proposed. Sequential covering is the kernel technique of them. To avoid testing all possible selectors, Entropy gain is used to select the best attribute in ID3. Constraint of the size of star was introduced in AQ and beam search was adopted in CN2. These methods speed up their induction algorithms but many good selectors are filtered out. In this work, we introduce a new induction algorithm that is based on enumeration of all possible selectors. Contrary to the previous works, we use pruning power to reduce irrelative selectors. But we can guarantee that no good selectors are filtered out. Comparing with other techniques, the experiment results demonstrate
that the rules produced by our induction algorithm have high consistency and simplicity.

An improved approach for the discovery of causal models via MML

Discovering a precise causal structure accurately reflecting the given data is one of the most essential tasks in the area of data mining and machine learning. One of the successful causal discovery approaches is the information-theoretic approach using the Minimum Message Length Principle[19]. This paper presents an improved and further experimental results of the MML discovery algorithm. We introduced a new encoding scheme for measuring the cost of describing the causal structure. Stiring function is also applied to further simplify the computational complexity and thus works more efficiently. The experimental results of the current version of the discovery system show that: (1) the current version is capable of discovering what discovered by previous system; (2) current system is capable of discovering more complicated causal models with large number of variables; (3) the new version works more efficiently compared with the previous version in terms of time complexity.

Improved support vector machine generalization using normalized input space

Data pre-processing always plays a key role in learning algorithm performance. In this research we consider data pre-processing by normalization for Support Vector Machines (SVMs). We examine the normalization affect across 112 classification problems with SVM using the rbf kernel. We observe a significant classification improvement due to normalization. Finally we suggest a rule based method to find when normalization is necessary for a specific classification problem. The best normalization method is also automatically selected by SVM itself.

A stable MCA learning algorithm

Minor component analysis (MCA) is an important statistical tool for signal processing and data analysis. Neural networks can be used to extract online minor component from input data. Compared with traditional algebraic  approaches, a neural network method has a lower computational complexity. Stability of neural networks learning algorithms is crucial to practical applications. In this paper, we propose a stable MCA neural networks learning algorithm, which has a more satisfactory numerical stability than some existing MCA algorithms. Dynamical behaviors of the proposed algorithm are analyzed via deterministic discrete time (DDT) method and the conditions are obtained to guarantee convergence. Simulations are carried out to illustrate the theoretical results achieved.

Exploiting geometry for improved hybrid AOA/TDOA-based localization

In this paper we examine the geometrically constrained optimization approach to localization with hybrid bearing (angle of arrival, AOA) and time difference of  arrival (TDOA) sensors. In particular, we formulate a constraint on the measurement errors which is then used along with constraint-based optimization tools in order to estimate the maximum likelihood values of the errors given an appropriate cost function. In particular we focus on deriving a localization algorithm for stationary target localization in the so-called adverse localization geometries where the relative positioning of the sensors and the target do not readily permit accurate or convergent localization using traditional approaches. We illustrate this point via simulation and we compare our approach to a number of different techniques that are discussed in the literature.