Parallel streaming signature EM-tree: A clustering algorithm for web scale applications

The proliferation of the web presents an unsolved problem of automatically analyzing billions of pages of natural language. We introduce a scalable algorithm that clusters hundreds of millions of web pages into hundreds of thousands of clusters. It does this on a single mid-range machine using efficient algorithms and compressed document representations. It is applied to two web-scale crawls covering tens of terabytes. ClueWeb09 and ClueWeb12 contain 500 and 733 million web pages and were clustered into 500,000 to 700,000 clusters. To the best of our knowledge, such fine grained clustering has not been previously demonstrated. Previous approaches clustered a sample that limits the maximum number of discoverable clusters. The proposed EM-tree algorithm uses the entire collection in clustering and produces several orders of magnitude more clusters than the existing algorithms. Fine grained clustering is necessary for meaningful clustering in massive collections where the number of distinct topics grows linearly with collection size. These fine-grained clusters show an improved cluster quality when assessed with two novel evaluations using ad hoc search relevance judgments and spam classifications for external validation. These evaluations solve the problem of assessing the quality of clusters where categorical labeling is unavailable and unfeasible.

Different approaches of applying single-objective binary genetic algorithm on the wind farm design

This paper discusses three different ways of applying the single-objective binary genetic algorithm into designing the wind farm. The introduction of different applications is through altering the binary encoding methods in GA codes. The first encoding method is the traditional one with fixed wind turbine positions. The second involves varying the initial positions from results of the first method, and it is achieved by using binary digits to represent the coordination of wind turbine on X or Y axis. The third is the mixing of the first encoding method with another one, which is by adding four more binary digits to represent one of the unavailable plots. The goal of this paper is to demonstrate how the single-objective binary algorithm can be applied and how the wind turbines are distributed under various conditions with best fitness. The main emphasis of discussion is focused on the scenario of wind direction varying from 0° to 45°. Results show that choosing the appropriate position of wind turbines is more significant than choosing the wind turbine numbers, considering that the former has a bigger influence on the whole farm fitness than the latter. And the farm has best performance of fitness values, farm efficiency, and total power with the direction between 20°to 30°.

Structural identification of a laboratory-based steel through truss cantilevered bridge with suspended span using a layered genetic algorithm with patternsearch step (GAPS)

Structural identification (St-Id) can be considered as the process of updating a finite element (FE) model of a structural system to match the measured response of the structure. This paper presents the St-Id of a laboratory-based steel through-truss cantilevered bridge with suspended span. There are a total of 600 degrees of freedom (DOFs) in the superstructure plus additional DOFs in the substructure. The St-Id of the bridge model used the modal parameters from a preliminary modal test in the objective function of a global optimisation technique using a layered genetic algorithm with patternsearch step (GAPS). Each layer of the St-Id process involved grouping of the structural parameters into a number of updating parameters and running parallel optimisations. The number of updating parameters was increased at each layer of the process. In order to accelerate the optimisation and ensure improved diversity within the population, a patternsearch step was applied to the fittest individuals at the end of each generation of the GA. The GAPS process was able to replicate the mode shapes for the first two lateral sway modes and the first vertical bending mode to a high degree of accuracy and, to a lesser degree, the mode shape of the first lateral bending mode. The mode shape and frequency of the torsional mode did not match very well. The frequencies of the first lateral bending mode, the first longitudinal mode and the first vertical mode matched very well. The frequency of the first sway mode was lower and that of the second sway mode was higher than the true values, indicating a possible problem with the FE model. Improvements to the model and the St-Id process will be presented at the upcoming conference and compared to the results presented in this paper. These improvements will include the use of multiple FE models in a multi-layered, multi-solution, GAPS St-Id approach.

Statistically assisted fluid registration algorithm - SAFIRA

In this paper, we develop and validate a new Statistically Assisted Fluid Registration Algorithm (SAFIRA) for brain images. A non-statistical version of this algorithm was first implemented in [2] and re-formulated using Lagrangian mechanics in [3]. Here we extend this algorithm to 3D: given 3D brain images from a population, vector fields and their corresponding deformation matrices are computed in a first round of registrations using the non-statistical implementation. Covariance matrices for both the deformation matrices and the vector fields are then obtained and incorporated (separately or jointly) in the regularizing (i.e., the non-conservative Lagrangian) terms, creating four versions of the algorithm. We evaluated the accuracy of each algorithm variant using the manually labeled LPBA40 dataset, which provides us with ground truth anatomical segmentations. We also compared the power of the different algorithms using tensor-based morphometry -a technique to analyze local volumetric differences in brain structure- applied to 46 3D brain scans from healthy monozygotic twins.

Using genetic algorithm in profile-based assignment of applications to virtual machines for greener data centers

The increase in data center dependent services has made energy optimization of data centers one of the most exigent challenges in today's Information Age. The necessity of green and energy-efficient measures is very high for reducing carbon footprint and exorbitant energy costs. However, inefficient application management of data centers results in high energy consumption and low resource utilization efficiency. Unfortunately, in most cases, deploying an energy-efficient application management solution inevitably degrades the resource utilization efficiency of the data centers. To address this problem, a Penalty-based Genetic Algorithm (GA) is presented in this paper to solve a defined profile-based application assignment problem whilst maintaining a trade-off between the power consumption performance and resource utilization performance. Case studies show that the penalty-based GA is highly scalable and provides 16% to 32% better solutions than a greedy algorithm.

A penalty-based genetic algorithm for the migration cost-aware virtual machine placement problem in cloud data centers

In the past few years, the virtual machine (VM) placement problem has been studied intensively and many algorithms for the VM placement problem have been proposed. However, those proposed VM placement algorithms have not been widely used in today's cloud data centers as they do not consider the migration cost from current VM placement to the new optimal VM placement. As a result, the gain from optimizing VM placement may be less than the loss of the migration cost from current VM placement to the new VM placement. To address this issue, this paper presents a penalty-based genetic algorithm (GA) for the VM placement problem that considers the migration cost in addition to the energy-consumption of the new VM placement and the total inter-VM traffic flow in the new VM placement. The GA has been implemented and evaluated by experiments, and the experimental results show that the GA outperforms two well known algorithms for the VM placement problem.

A Strength Pareto Evolutionary Algorithm for live migration of multiple interdependent virtual machines in data centers

Although live VM migration has been intensively studied, the problem of live migration of multiple interdependent VMs has hardly been investigated. The most important problem in the live migration of multiple interdependent VMs is how to schedule VM migrations as the schedule will directly affect the total migration time and the total downtime of those VMs. Aiming at minimizing both the total migration time and the total downtime simultaneously, this paper presents a Strength Pareto Evolutionary Algorithm 2 (SPEA2) for the multi-VM migration scheduling problem. The SPEA2 has been evaluated by experiments, and the experimental results show that the SPEA2 can generate a set of VM migration schedules with a shorter total migration time and a shorter total downtime than an existing genetic algorithm, namely Random Key Genetic Algorithm (RKGA). This paper also studies the scalability of the SPEA2.

An algorithm for constructing Hjelmslev planes

Projective Hjelmslev planes and affine Hjelmslev planes are generalisations of projective planes and affine planes. We present an algorithm for constructing projective Hjelmslev planes and affine Hjelmslev planes that uses projective planes, affine planes and orthogonal arrays. We show that all 2-uniform projective Hjelmslev planes, and all 2-uniform affine Hjelmslev planes can be constructed in this way. As a corollary it is shown that all $2$-uniform affine Hjelmslev planes are sub-geometries of $2$-uniform projective Hjelmslev planes.

A data mining approach for fault diagnosis: An application of anomaly detection algorithm

Rolling-element bearing failures are the most frequent problems in rotating machinery, which can be catastrophic and cause major downtime. Hence, providing advance failure warning and precise fault detection in such components are pivotal and cost-effective. The vast majority of past research has focused on signal processing and spectral analysis for fault diagnostics in rotating components. In this study, a data mining approach using a machine learning technique called anomaly detection (AD) is presented. This method employs classification techniques to discriminate between defect examples. Two features, kurtosis and Non-Gaussianity Score (NGS), are extracted to develop anomaly detection algorithms. The performance of the developed algorithms was examined through real data from a test to failure bearing. Finally, the application of anomaly detection is compared with one of the popular methods called Support Vector Machine (SVM) to investigate the sensitivity and accuracy of this approach and its ability to detect the anomalies in early stages.

One practical algorithm for both stochastic and adversarial bandits

We present an algorithm for multiarmed bandits that achieves almost optimal performance in both stochastic and adversarial regimes without prior knowledge about the nature of the environment. Our algorithm is based on augmentation of the EXP3 algorithm with a new control lever in the form of exploration parameters that are tailored individually for each arm. The algorithm simultaneously applies the “old” control lever, the learning rate, to control the regret in the adversarial regime and the new control lever to detect and exploit gaps between the arm losses. This secures problem-dependent “logarithmic” regret when gaps are present without compromising on the worst-case performance guarantee in the adversarial regime. We show that the algorithm can exploit both the usual expected gaps between the arm losses in the stochastic regime and deterministic gaps between the arm losses in the adversarial regime. The algorithm retains “logarithmic” regret guarantee in the stochastic regime even when some observations are contaminated by an adversary, as long as on average the contamination does not reduce the gap by more than a half. Our results for the stochastic regime are supported by experimental validation.

Predicting tours and probabilistic simulation for BKZ lattice reduction algorithm

We investigate the terminating concept of BKZ reduction first introduced by Hanrot et al. [Crypto'11] and make extensive experiments to predict the number of tours necessary to obtain the best possible trade off between reduction time and quality. Then, we improve Buchmann and Lindner's result [Indocrypt'09] to find sub-lattice collision in SWIFFT. We illustrate that further improvement in time is possible through special setting of SWIFFT parameters and also through the combination of different reduction parameters adaptively. Our contribution also include a probabilistic simulation approach top-up deterministic simulation described by Chen and Nguyen [Asiacrypt'11] that can able to predict the Gram-Schmidt norms more accurately for large block sizes.

A robust approach to design a single facility layout plan in dynamic manufacturing environments using a permutation-based genetic algorithm

During the past few decades, developing efficient methods to solve dynamic facility layout problems has been focused on significantly by practitioners and researchers. More specifically meta-heuristic algorithms, especially genetic algorithm, have been proven to be increasingly helpful to generate sub-optimal solutions for large-scale dynamic facility layout problems. Nevertheless, the uncertainty of the manufacturing factors in addition to the scale of the layout problem calls for a mixed genetic algorithm–robust approach that could provide a single unlimited layout design. The present research aims to devise a customized permutation-based robust genetic algorithm in dynamic manufacturing environments that is expected to be generating a unique robust layout for all the manufacturing periods. The numerical outcomes of the proposed robust genetic algorithm indicate significant cost improvements compared to the conventional genetic algorithm methods and a selective number of other heuristic and meta-heuristic techniques.

Algorithm of medication review in frail older people: Focus on minimizing the use of high-risk medications

Aim Frail older people typically suffer several chronic diseases, receive multiple medications and are more likely to be institutionalized in residential aged care facilities. In such patients, optimizing prescribing and avoiding use of high-risk medications might prevent adverse events. The present study aimed to develop a pragmatic, easily applied algorithm for medication review to help clinicians identify and discontinue potentially inappropriate high-risk medications. Methods The literature was searched for robust evidence of the association of adverse effects related to potentially inappropriate medications in older patients to identify high-risk medications. Prior research into the cessation of potentially inappropriate medications in older patients in different settings was synthesized into a four-step algorithm for incorporation into clinical assessment protocols for patients, particularly those in residential aged care facilities. Results The algorithm comprises several steps leading to individualized prescribing recommendations: (i) identify a high-risk medication; (ii) ascertain the current indications for the medication and assess their validity; (iii) assess if the drug is providing ongoing symptomatic benefit; and (iv) consider withdrawing, altering or continuing medications. Decision support resources were developed to complement the algorithm in ensuring a systematic and patient-centered approach to medication discontinuation. These include a comprehensive list of high-risk medications and the reasons for inappropriateness, lists of alternative treatments, and suggested medication withdrawal protocols. Conclusions The algorithm captures a range of different clinical scenarios in relation to potentially inappropriate medications, and offers an evidence-based approach to identifying and, if appropriate, discontinuing such medications. Studies are required to evaluate algorithm effects on prescribing decisions and patient outcomes.

Design and flight testing of a bio-inspired plume tracking algorithm for unmanned serial vehicles

There is an increased interest on the use of UAVs for environmental research such as tracking bush fires, volcanic eruptions, chemical accidents or pollution sources. The aim of this paper is to describe the theory and results of a bio-inspired plume tracking algorithm. A method for generating sparse plumes in a virtual environment was also developed. Results indicated the ability of the algorithms to track plumes in 2D and 3D. The system has been tested with hardware in the loop (HIL) simulations and in flight using a CO2 gas sensor mounted to a multi-rotor UAV. The UAV is controlled by the plume tracking algorithm running on the ground control station (GCS).

FreeS: A fast algorithm to discover frequent free subtrees using a novel canonical form

Web data can often be represented in free tree form; however, free tree mining methods seldom exist. In this paper, a computationally fast algorithm FreeS is presented to discover all frequently occurring free subtrees in a database of labelled free trees. FreeS is designed using an optimal canonical form, BOCF that can uniquely represent free trees even during the presence of isomorphism. To avoid enumeration of false positive candidates, it utilises the enumeration approach based on a tree-structure guided scheme. This paper presents lemmas that introduce conditions to conform the generation of free tree candidates during enumeration. Empirical study using both real and synthetic datasets shows that FreeS is scalable and significantly outperforms (i.e. few orders of magnitude faster than) the state-of-the-art frequent free tree mining algorithms, HybridTreeMiner and FreeTreeMiner.