Search

Fog computing is an increasingly popular method with which to process the huge amount of data generated by the Internet of Things (IoT) devices and applications at the edge-level, using the heterogeneous autonomous end-devices of the participating users. To meet the requirements of the IoT and time-sensitive applications, a Fog computing platform needs to select appropriate resources, the availability of which can be guaranteed during the execution of the application. For the proper selection of resources, the platform must be able to predict future availability. Hence, a proper resource availability model which provides knowledge about the future availability of resources in the Fog computing environment is required. However, designing an efficient resource availability model, in a highly distributed and mobile environment like the Fog, is a complex task due to the multidimensional characteristics of Fog devices, such as mobility, lack of centralised control, limited resources, and being battery powered. Existing resource availability models did not consider all the characteristics of a real Fog environment. Therefore, this study aims to provide a generic continuous-time Markov chain (CTMC), based resource availability model for Fog computing environments. The applicability of the model is shown by integrating the model input with the nearest-location best fit (NLBF) and Best-Fit resource selection policies.

With the increasing number of Internet of Things (IoT) devices, the volume and variety of data being generated by these devices are increasing rapidly. Cloud computing cannot process this data due to its high latency and scalability. In order to process this data in less time, fog computing has evolved as an extension to Cloud computing. In a fog computing environment, a resource monitoring service plays a vital role in providing advanced services, such as scheduling, scaling and migration. Most of the research in fog computing has assumed that a resource monitoring service is already available. Conventional methods proposed for other distributed systems may not be suitable due to the unique features of a fog environment. To improve the overall performance of fog computing and to optimise resource usage, effective resource monitoring techniques are required. Hence, we propose a support and confidence based (SCB) technique which optimises the resource usage in the resource monitoring service. The performance of our proposed system is evaluated by examining a real-time traffic use case in a fog emulator with synthetic data. The experimental results obtained from the fog emulator show that the proposed technique consumes 19 percent lesser resources compared with the existing technique.

Objective Instruments rating risk of harm to self and others are widely used in inpatient forensic psychiatry settings. A potential alternate or supplementary means of risk prediction is from the automated analysis of case notes in Electronic Health Records (EHRs) using Natural Language Processing (NLP). This exploratory study rated presence or absence and frequency of words in a forensic EHR dataset, comparing four reference dictionaries. Seven machine learning algorithms and different time periods of EHR analysis were used to probe which dictionary and which time period were most predictive of risk assessment scores on validated instruments. Materials and methods The EHR dataset comprised de-identified forensic inpatient notes from the Wilfred Lopes Centre in Tasmania. The data comprised unstructured free-text case note entries and serial ratings of three risk assessment scales: Historical Clinical Risk Management-20 (HCR-20), Short-Term Assessment of Risk and Treatability (START) and Dynamic Appraisal of Situational Aggression (DASA). Four NLP dictionary word lists were selected: 6865 mental health symptom words from the Unified Medical Language System (UMLS), 455 DSM-IV diagnoses from UMLS repository, 6790 English positive and negative sentiment words, and 1837 high frequency words from the Corpus of Contemporary American English (COCA). Seven machine learning methods Bagging, J48, Jrip, Logistic Model Trees (LMT), Logistic Regression, Linear Regression and Support Vector Machine (SVM) were used to identify the combination of dictionaries and algorithms that best predicted risk assessment scores. Results The most accurate prediction was attained on the DASA dataset using the sentiment dictionary and the LMT and SVM algorithms. Conclusions NLP, used in conjunction with NLP dictionaries and machine learning, predicted risk ratings on the HCR-20, START, and DASA, based on EHR content. Further research is required to ascertain the utility of NLP approaches in predicting endpoints of actual self-harm, harm to others or victimisation.

Recommender systems have become an integral part of many social networks and extract knowledge from a user’s personal and sensitive data both explicitly, with the user’s knowledge, and implicitly. This trend has created major privacy concerns as users are mostly unaware of what data and how much data is being used and how securely it is used. In this context, several works have been done to address privacy concerns for usage in online social network data and by recommender systems. This paper surveys the main privacy concerns, measurements and privacy-preserving techniques used in large-scale online social networks and recommender systems. It is based on historical works on security, privacy-preserving, statistical modeling, and datasets to provide an overview of the technical difficulties and problems associated with privacy preserving in online social networks.

This study compares an adaptive and a nonadaptive representation for finding long walks on obstructed grids. This process models adaption of a simple plant to an environment where the plant's ability to grow is impeded by obstructions such as resource poor areas like bare rock. The intent of the adaptive representation is to model the biological phenomenon of phenotypic plasticity in which gene regulation is at least partially in response to environmental cues, in this case the obstructions. The adaptive representation is found to have a substantial advantage, with the greatest level of advantage at intermediate levels of obstruction. Agents are asked to solve multiple problem instances simultaneously (i.e. using the same chromosome). The advantage of the adaptive representation is also found to be higher when more boards are used in fitness evaluation.

This study uses a real-coded representation to encode discrete strategies for playing a simple grid based game. This representation is able to adapt itself on the fly to the local situation in the game. This adaptability makes the representation particularly suitable for finding good play strategies which, in turn, permit us to explore biases in the play strategies and even to compare instances of the game for difficulty. Variability in the best results achieved by the solver can be used to gauge difficulty, while the shape of the distribution of best results can indicate how interesting an instance is. Results indicate that the variety and combination of affordances produce instances of the game with varying degrees of anticipated difficulty and interestingness, and confirm that the solver can be used to evaluate the quality of different affordance combinations for producing good game instances. Design principles may also be discovered through post hoc analysis of instances with high or low average best score.

The goal of exploration to produce diverse search points throughout the search space can be countered by the goal of selection to focus search around the fittest current solution(s). In the limit, if all exploratory search points are rejected by selection, then the behaviour of the metaheuristic will be equivalent to one which performs no exploration at all (e.g. hill climbing). The effects of selection on exploration are clearly important, but our review of the literature indicates limited coverage. To address this deficit, we introduce new experiments which can specifically highlight the occurrence of “failed exploration” and its effects through selection that can trap a metaheuristic in a less promising part of the search space. We subsequently propose new lines of research to reduce the effects of selection and failed exploration which we believe are distinctly different from traditional lines of research to increase (pre-selection) exploration.

Automatic recording and analysis of bird calls is becoming an important way to understand changes in bird populations and assess environmental health. An issue currently proving problematic with the automatic analysis of bird recordings is interference from noise that can mask vocalizations of interest. As such, noise reduction can greatly increase the accuracy of automatic analyses and reduce processing work for subsequent steps in bioacoustics analyses. However, only limited work has been done in the context of bird recordings. Most semiautomatic methods either manually apply sound enhancement methods available in audio processing systems such as SoX and Audacity or apply preliminary filters such as low- and high-pass filters. These methods are insufficient both in terms of how generically they can be applied and their integration with automatic systems that need to process large amounts of data. Some other work applied more sophisticated denoising methods or combinations of different methods such as minimum mean square error short-time spectral amplitude estimator (MMSE STSA) and spectral subtraction for other species such as anurans. However, their effectiveness is not tested on bird recordings. In this paper, we analyze the applicability of the MMSE STSA algorithm to remove noise from environmental recordings containing bird sounds, particularly focusing on its quality and processing time. The experimental evaluation using real data clearly shows that MMSE STSA can reduce noise with similar effectiveness [using objective metrics such as predicted signal quality (SIG)] to a previously recommended wavelet-transform-based denoising technique while executing between approximately 5–300 times faster depending on the audio files tested.

Users in online social networking sites unknowingly disclose their sensitive information that aggravate the social and financial risks. Hence, to prevent the information loss and privacy exposure, users need to find ways to quantify their privacy level based on their online social network data. Current studies that focus on measuring the privacy risk and disclosure consider only a single source of data, neglecting the fact that users, in general, can have multiple social network accounts disclosing different sensitive information. In this paper, we investigate an approach that can help social media users to measure their privacy disclosure score (PDS) based on the information shared across multiple social networking sites. In particular, we identify the main factors that have impact on users privacy, namely, sensitivity and visibility, to obtain the final disclosure score for each user. By applying the statistical and fuzzy systems, we can specify the potential information loss for a user by using obtained PDS. Our evaluation results with real social media data show that our method can provide a better estimation of privacy disclosure score for users having presence in multiple online social networks.

Online social network users share their information in different social sites to establish connections with individuals with whom they want to be a friend. While users share all their information to connect to other individuals, they need to hide the information that can bring about privacy risks for them. As user participation in social networking sites rises, the possibility of sharing information with unknown users increases, and the probability of privacy breaches for the user mounts. This work addresses the challenges of sharing information in a safe manner with unknown individuals. Currently, there are a number of available methods for preserving privacy in order to friending (the act of adding someone as a friend), but they only consider a single source of data and are more focused on users&rsquo, security rather than privacy. Consequently, a privacy-preserving friending mechanism should be considered for information shared in multiple online social network sites. In this paper, we propose a new privacy-preserving friending method that helps users decide what to share with other individuals with the reduced risk of being exploited or re-identified. In this regard, the first step is to calculate the sensitivity score for individuals using Bernstein&rsquo, s polynomial theorem to understand what sort of information can influence a user&rsquo, s privacy. Next, a new model is applied to anonymise the data of users who participate in multiple social networks. Anonymisation helps to understand to what extent a piece of information can be shared, which allows information sharing with reduced risks in privacy. Evaluation indicates that measuring the sensitivity of information besides anonymisation provides a more accurate outcome for the purpose of friending, in a computationally efficient manner.

The use of water resources for agricultural purposes, particularly in arid and semi-arid regions, is a matter of increasing concern across the world. Optimisation techniques can play an important role in improving the allocation of land to different crops, based on a utility function (such as net revenue) and the water resources needed to support these. Recent work proposed a model formulation for an agricultural region in the Murrumbidgee Irrigation Area of the Murray-Darling River basin in Australia, and found that the well-known NSGA-II technique could produce sensible crop mixes while preserving ground and surface water for environmental purposes. In the present study we apply Differential Evolution using two different solution representations, one of which explores the restricted space in which no land is left fallow. The results improve on those of the prior NSGA-II and demonstrate that a combination of solution representations allows Differential Evolution to more thoroughly explore the multiobjective space of profit versus environment.

One of the common forms of data which is shared by online social media users is free-text formats including comments, posts, blogs and tweets. While users mostly share this unstructured data with their preferred social groups, this textual data may contain sensitive information such as their political or religious views, job details, their opinions and emotions and so on. Hence, sharing this unstructured data can escalate privacy risks and concerns for social media users. Analyses the privacy of unstructured data occurred from textual information comes with difficulties as understanding the calculation metrics are challenging. Although there are various studies on privacy evaluation from the extracted structured information from unstructured data, there are limited privacy scoring methods concentrating on the views of the individuals and cannot satisfy the privacy scoring of shared unstructured data in social networks appropriately. Here, in this paper, we propose an automated fuzzy-based model that can extract the privacy-related features as well as the related shared structured data and measure and warn users regarding the textual data privacy risks they have shared in online social platforms. The proposed model can facilitate mitigation actions for users’ free-format texts shared in various social networks. The evaluation of the study indicates that the proposed model can measure the users’ privacy risk in a more accurate manner compared with previously proposed methods and available commercialised software in the domain.

Optimisation problems usually take the form of having a single or multiple objectives with a set of constraints. The model itself concerns a single problem for which the best possible solution is sought. Problems are usually static in the sense that they do not consider changes over time in a cumulative manner. Dynamic optimisation problems to incorporate changes. However, these are memoryless in that the problem description changes and a new problem is solved - but with little reference to any previous information. In this paper, a temporally augmented version of a water management problem which allows farmers to plan over long time horizons is introduced. A climate change projection model is used to predict both rainfall and temperature for the Murrumbidgee Irrigation Area in Australia for up to 50 years into the future. Three representative decades are extracted from the climate change model to create the temporal data sets. The results confirm the utility of the temporal approach and show, for the case study area, that crops that can feasibly and sustainably be grown will be a lot fewer than the present day in the challenging water-reduced conditions of the future.

The automated design of meander line RFID antennas is a discrete self-avoiding walk (SAW) problem for which efficiency is to be maximized while resonant frequency is to be minimized. This work presents a novel exploration of how discrete local search may be incorporated into a continuous solver such as differential evolution (DE). A prior DE algorithm for this problem that incorporates an adaptive solution encoding and a bias favoring antennas with low resonant frequency is extended by the addition of the backbite local search operator and a variety of schemes for reintroducing modified designs into the DE population. The algorithm is extremely competitive with an existing ACO approach and the technique is transferable to other SAW problems and other continuous solvers. The findings indicate that careful reintegration of discrete local search results into the continuous population is necessary for effective performance.

Recording and analysing environmental audio recordings has become a common approach for monitoring the environment. This has several advantages over other approaches, such as reducing costs by avoiding the need for experts to be present in the area of interest. A current problem with performing analyses of environmental recordings is interference from noise that can mask vocalisations of interest. This makes detecting these vocalisations more difficult and can require additional resources. While some work has been done to remove stationary noise from environmental recordings, there has been little effort to remove noise from non-stationary sources, such as rain, wind, engines, and animal vocalisations that are not of interest. This work addresses the challenge of filtering noise from rain and cicada choruses from recordings containing bird sound. The use of acoustic indices and Mel Frequency Cepstral Coefficients (MFCCs) with machine learning classifiers is investigated to find the most effective filters. Hyperparameters for several classification approaches are investigated to fine tune models to achieve the best results. The approach used enables users to set thresholds to increase or decrease the sensitivity of classification, based on the prediction probability outputted by classifiers. A novel approach to remove cicada choruses using bandpass filters is also proposed. A threshold-based approach (Multi-Layer Perceptron with Acoustic Indices and MFCCs) for rain detection is derived which achieves an AUC of 0.9911 and is more accurate than existing approaches when set to the same sensitivities. Cicada choruses are classified in the training set used with 100% accuracy using 10-fold cross-validation using a Support Vector Machine (SVM) classifier with MFCCs. The cicada filtering approach greatly increased the median signal to noise ratios of affected recordings from 0.53 for unfiltered audio to 1.86 to audio filtered by both the cicada filter and a common stationary noise filter.

In this work, we examine the problem of efficiently preprocessing and denoising high volume environmental acoustic data, which is a necessary step in many bird monitoring tasks. Preprocessing is typically made up of multiple steps which are considered separately from each other. These are often resource intensive, particularly because the volume of data involved is high. We focus on addressing two challenges within this problem: how to combine existing preprocessing tasks while maximising the effectiveness of each step, and how to process this pipeline quickly and efficiently, so that it can be used to process high volumes of acoustic data. We describe a distributed system designed specifically for this problem, utilising a master-slave model with data parallelisation. By investigating the impact of individual preprocessing tasks on each other, and their execution times, we determine an efficient and accurate order for preprocessing tasks within the distributed system. We find that, using a single core, our pipeline executes 1.40 times faster compared to manually executing all preprocessing tasks. We then apply our pipeline in the distributed system and evaluate its performance. We find that our system is capable of preprocessing bird acoustic recordings at a rate of 174.8 seconds of audio per second of real time with 32 cores over 8 virtual machines, which is 21.76 times faster than a serial process.

Exploration and exploitation are analyzed in Particle Swarm Optimization (PSO) through a set of experiments that make new measurements of these key features. Compared to analyses on diversity and particle trajectories, which focus on particle motions and their potential to achieve exploration and exploitation, our analysis also focuses on the pbest positions that reflect the actual levels of exploration and exploitation that have been achieved by PSO. A key contribution of this paper is a clear criterion for when restarting particles can be expected to be a useful strategy in PSO.

This study is the second using real-coded representation for problems usually solved with a discrete coding. The adaptive generative representation is able to adapt itself on the fly to prior parts of the construction of an object as it assembles it. In the initial study the ability of the representation to take user supplied or problem supplied biases that change its behavior was demonstrated but not explored. In this study the bias is used to change the way evolution explores a fitness landscape for both an RFID antenna design problem and small instances of the traveling salesman problem. Addition of a bias to two different generative representations promotes the evolution of longer antenna designs (a heuristic objective associated with good antennas) while leading the algorithm to generate designs with distinctive shape characteristics. For the traveling salesman, a simple inverse-distance bias for the adaptive generative representation causes a large improvement in performance over a random key representation in 99 of 100 instances studied.

Many application areas such as biodiversity monitoring and speech recognition produce several gigabytes of audio data per day, think of weeks, months of accumulated data available to analyse. However, current methods and frameworks can only process a few megabytes at a time. Moreover, these methods require a lot of manual processing before they can give relevant results for biodiversity researchers. In this paper, we present a novel scalable framework called B2P2 for handling large audio files and processing them in an efficient manner. The proposed framework can process and handle large audio data that can indicate whether there are sounds of potential interest throughout large audio recordings. We implemented the proposed framework using BigData platforms such as Spark and HDFS. Experimental results demonstrate an increase in speed-up and performance of processing audio files when increasing nodes and data sizes on cloud computing.

This study introduces a novel generative representation that is able to modify its expression in response to admissibility constraints that unfold as solutions are generated. The effect is that this self-adaptation in expression makes many inadmissible structures impossible to encode. The resulting reduction in the effective size of the search space yields performance increases amounting to several orders of magnitude for some problems. In addition to defining and exploring the capabilities of the self-adaptive representation, a technique for biasing its expression with numerical weights that strongly influences which optima are located is introduced. This both permits enhancement of optima with desirable properties and permits the inclusion of domain knowledge to improve performance. The test problems used are the self-avoiding walk problem, a surrogate for RFID tag antenna design, and the Towers of Hanoi problem.

With the globalisation of education, students may undertake higher education courses anywhere in the world. Yet there is variation between different universities’ offerings. Even though web search engines can help one to locate potentially similar courses or subjects offered by different universities, judging the degree of similarity between each of them is currently a manual process in which a student or staff member has to go through subject/unit descriptions within a course to understand the different topics taught. In this paper, we study the application of text mining to evaluate the similarity or overlap between different units and propose a system that can help students and staff to make these judgements. The unit or course descriptions are generally short, containing 100–200 words, and exhibit very wide diversity in the ways they are written. Experimental results using data from Australian and international universities demonstrate the accuracy of the proposed system in calculating the similarity between different computing units.