Description:
Diagnostic radiology struggles to maintain high interpretation accuracy. Retrieval of past similar cases would help the inexperienced radiologist in the interpretation process. Character n-gram model has been effective in text retrieval context in languages such as Chinese where there are no clear word boundaries. We propose the use of visual character n-gram model for representation of image for classification and retrieval purposes. Regions of interests in mammographic images are represented with the character n-gram features. These features are then used as input to back-propagation neural network for classification of regions into normal and abnormal categories. Experiments on miniMIAS database show that character n-gram features are useful in classifying the regions into normal and abnormal categories. Promising classification accuracies are observed (83.33%) for fatty background tissue warranting further investigation. We argue that Classifying regions of interests would reduce the number of comparisons necessary for finding similar images from the database and hence would reduce the time required for retrieval of past similar cases.
Description:
Diagnostic radiology struggles to maintain high interpretation accuracy. Retrieval of past similar cases would help the inexperienced radiologist in the interpretation process. Character n-gram model has been effective in text retrieval context in languages such as Chinese where there are no clear word boundaries. We propose the use of visual character n-gram model for representation of image for classification and retrieval purposes. Regions of interests in mammographic images are represented with the character n-gram features. These features are then used as input to back-propagation neural network for classification of regions into normal and abnormal categories. Experiments on miniMIAS database show that character n-gram features are useful in classifying the regions into normal and abnormal categories. Promising classification accuracies are observed (83.33%) for fatty background tissue warranting further investigation. We argue that Classifying regions of interests would reduce the number of comparisons necessary for finding similar images from the database and hence would reduce the time required for retrieval of past similar cases.
Description:
Cloud computing delivers computing as a utility to users worldwide. A consequence of this model is that cloud data centres have high deployment and operational costs, as well as significant carbon footprints for the environment. We need to develop Green Cloud Computing (GCC) solutions that reduce these deployment and operational costs and thus save energy and reduce adverse environmental impacts. In order to achieve this objective, a thorough understanding of the energy consumption patterns in complex Cloud environments is needed. We present a new energy consumption model and associated analysis tool for Cloud computing environments. We measure energy consumption in Cloud environments based on different runtime tasks. Empirical analysis of the correlation of energy consumption and Cloud data and computational tasks, as well as system performance, will be investigated based on our energy consumption model and analysis tool. Our research results can be integrated into Cloud systems to monitor energy consumption and support static or dynamic system-level optimisation.
Description:
Spectrum allocation scheme in cognitive radio networks (CRNs) becomes complex when multiple CR users concomitantly need to be allocated new and suitable bands once the primary user returns. Most existing schemes focus on the gain of individual users, ignoring the effect of an allocation on other users and rely on the 'periodic sensing and transmission' cycle which reduces spectrum utilization. This paper introduces a scheme that exploits collaboration among users to detect PU's return which relieves active CR users from the sensing task, and thereby improves spectrum utilization. It defines a Capacity of Service (CoS) metric based on the optimal sensing parameters which measures the suitability of a band for each contending user and takes into consideration the impact of allocating a particular band on other band seeking users. The proposed scheme significantly improves capacity of service, reduces interference loss and collision, and hence, enhances dynamic spectrum access capabilities. (C) 2015 Elsevier Ltd. All rights reserved.
Description:
Multiple-point defects and abraded surfaces in rotary machinery induce complex vibration signatures, and have a tendency to mislead defect diagnosis models. A challenging problem in machine defect diagnosis is to model and study defect signature dynamics in the case of multiple-point defects and surface abrasion. In this study, a multiple-point defect model (MPDM) that characterizes the dynamics of n-point bearing defects is proposed. MPDM is further extended to model degradation in a rotating machine as a special case of multiple-point defects. Analytical and experimental results for multiple-point defects and abrasions show that the location of the fundamental defect frequency shifts depending upon the relative location of the defects and width of the abrasive region. This variation in the defect frequency results in a degradation of the defect detection accuracy of the defect diagnostic model. Based on envelope detection analysis, a modification in existing defect diagnostic models is recommended to nullify the impact of multiple-point defects, and general abrasion in machine components.
Description:
Mining interesting knowledge from the huge amount of data gathered from WSNs is a challenge. Works reported in literature use support metric-based sensor association rules which employ the occurrence frequency of patterns as criteria. However, consideration of the binary frequency of a pattern is not a sufficient indicator for finding meaningful patterns because it only reflects the number of epochs which contain that pattern in the dataset. The share measure of sensorsets could discover useful knowledge about trigger values associated with a sensor. Here, we propose a new type of behavioral pattern called share-frequent sensor patterns (SFSPs) by considering the non-binary frequency values of sensors in epochs. SFSPs can find a correlation among a set of sensors and hence can improve the performance of WSNs in a resource management process. In this paper, a share-frequent sensor pattern tree (ShrFSP-Tree) has been proposed to facilitate a pattern growth mining technique to discover SFSPs from WSN data. We also present a parallel and distributed method where the ShrFSP-Tree is enhanced into PShrFSP-Tree and its performance is investigated for both homogeneous and heterogeneous systems. Results show that our method is time and memory efficient in finding SFSPs than the existing most efficient algorithms.
Description:
Cloud computing delivers IT solutions as a utility to users. A common economic objective for both cloud consumers and providers is to minimise their total deployment and operational costs while achieving satisfactory system performance to meet Service Level Agreements (SLAs). Therefore, the trade-off between cost and system performance is needed to be managed to achieve the best cost effectiveness. The PhD project presented in this paper provides an efficient and accurate method to evaluate the trade-off between cost and system performance by proposing a performance testing tool for Cloud systems. This tool can accommodate different Cloud system architectures and adopt different workload and cost models of Cloud systems during the trade-off evaluation process.
Description:
The 3D Morphable models of the human face have prepared myriad of applications in computer vision, human computer interaction and security surveillances. However, due to the variation in size, complexity of training data set, the landmark mapping, the representation in real time and rendering or synthesis of images in three dimensional is limited. In this paper, we extend the approach of the photometric stereo and provide the human face reconstruction in three dimensional. The proposed method consists of two steps. First it automatically detects the face and segment the iris along with statistical features of pupil location in it. Secondly it provides the selection of minimum six features and where iris process to generate the 3D face. In compare with existing methods our approach provides the automation which produces more better and efficient results in contrast to the manual methods.
Description:
The 3D Morphable models of the human face have prepared myriad of applications in computer vision, human computer interaction and security surveillances. However, due to the variation in size, complexity of training data set, the landmark mapping, the representation in real time and rendering or synthesis of images in three dimensional is limited. In this paper, we extend the approach of the photometric stereo and provide the human face reconstruction in three dimensional. The proposed method consists of two steps. First it automatically detects the face and segment the iris along with statistical features of pupil location in it. Secondly it provides the selection of minimum six features and where iris process to generate the 3D face. In compare with existing methods our approach provides the automation which produces more better and efficient results in contrast to the manual methods.
Description:
While microarrays make it feasible to rapidly investigate many complex biological problems, their multistep fabrication has the proclivity for error at every stage. The standard tactic has been to either ignore or regard erroneous gene readings as missing values, though this assumption can exert a major influence upon postgenomic knowledge discovery methods like gene selection and gene regulatory network (GRN) reconstruction. This has been the catalyst for a raft of new flexible imputation algorithms including local least square impute and the recent heuristic collateral missing value imputation, which exploit the biological transactional behaviour of functionally correlated genes to afford accurate missing value estimation. This paper examines the influence of missing value imputation techniques upon postgenomic knowledge inference methods with results for various algorithms consistently corroborating that instead of ignoring missing values, recycling microarray data by flexible and robust imputation can provide substantial performance benefits for subsequent downstream procedures
Description:
While microarrays make it feasible to rapidly investigate many complex biological problems, their multistep fabrication has the proclivity for error at every stage. The standard tactic has been to either ignore or regard erroneous gene readings as missing values, though this assumption can exert a major influence upon postgenomic knowledge discovery methods like gene selection and gene regulatory network (GRN) reconstruction. This has been the catalyst for a raft of new flexible imputation algorithms including local least square impute and the recent heuristic collateral missing value imputation, which exploit the biological transactional behaviour of functionally correlated genes to afford accurate missing value estimation. This paper examines the influence of missing value imputation techniques upon postgenomic knowledge inference methods with results for various algorithms consistently corroborating that instead of ignoring missing values, recycling microarray data by flexible and robust imputation can provide substantial performance benefits for subsequent downstream procedures
Description:
Trustworthiness in an online environment is essential because individuals and organizations can easily be misled by false and malicious information receiving from untrustworthy users. Though existing methods assess users' trustworthiness by exploiting Twitter account properties, their efficacy is inadequate because of Twitter's restriction on profile and tweet size, the existence of missing or insufficient profiles, and ease to create fake accounts or relationships to pretend as trustworthy. In this paper, we present a holistic approach by exploiting ideas perceived from real-world organizations for trust estimation along with available Twitter information. Users' trustworthiness is determined by considering their credentials, recommendation from referees and the quality of the information in their Twitter accounts and tweets. We establish the feasibility of our approach analytically and further devise a multi-objective cost function for the A
Description:
Trustworthiness in an online environment is essential because individuals and organizations can easily be misled by false and malicious information receiving from untrustworthy users. Though existing methods assess users' trustworthiness by exploiting Twitter account properties, their efficacy is inadequate because of Twitter's restriction on profile and tweet size, the existence of missing or insufficient profiles, and ease to create fake accounts or relationships to pretend as trustworthy. In this paper, we present a holistic approach by exploiting ideas perceived from real-world organizations for trust estimation along with available Twitter information. Users' trustworthiness is determined by considering their credentials, recommendation from referees and the quality of the information in their Twitter accounts and tweets. We establish the feasibility of our approach analytically and further devise a multi-objective cost function for the A
Description:
High density of coexisting networks in the Industrial, Scientific and Medical (ISM) band leads to static and self interferences among different communication entities. The inevitability of these interferences demands for interference avoidance schemes to ensure reliability of network operations. This paper proposes a novel Diversified Adaptive Frequency Rolling (DAFR) technique for frequency hopping in Bluetooth piconets. DAFR employs intelligent hopping procedures in order to mitigate self interferences, weeds out the static interferer efficiently and ensures sufficient frequency diversity. We compare the performance of our proposed technique with the widely used existing frequency hopping techniques, namely, Adaptive Frequency Hopping (AFH) and Adaptive Frequency Rolling (AFR). Simulation studies validate the significant improvement in goodput and hopping diversity of our scheme compared to other schemes and demonstrate its potential benefit in real world deployment.
Description:
Mining of sensor data for useful knowledge extraction is a very challenging task. Existing works generate sensor association rules using occurrence frequency of patterns to extract the knowledge. These techniques often generate huge number of rules, most of which are non-informative or fail to reflect true correlation among sensor data. In this paper, we propose a new type of behavioral pattern called associated sensor patterns which capture association-like co-occurrences as well as temporal correlations which are linked with such co-occurrences. To capture such patterns a compact tree structure, called associated sensor pattern tree (ASP-tree) and a mining algorithm (ASP) are proposed which use pattern growth-based approach to generate all associated patterns with only one scan over dataset. Moreover, when data stream flows through, old information may lose significance for the current time. To capture significance of recent data, ASP-tree is further enhanced to SWASP-tree by adopting sliding observation window and updating the tree structure accordingly. Finally, window size is made dynamically adaptive to ensure efficient resource usage. Different characteristics of the proposed techniques and their computational complexity are presented. Experimental results show that our approach is very efficient in discovering associated sensor patterns and outperforms existing techniques.
Description:
There has been a growing interest in implementing image search engine at the semantic level. However, most existing practical systems including popular commercial image search engines like Google and Yahoo! are either text-based or a simple hybrid of texts and visual features. This paper proposes a novel image search system based on automatic image annotation. We develop a technology which learns semantic image concepts from image contents and transforms unstructured images into textual documents, so that images are indexed and retrieved in the same way as textual documents. Existing database management systems can be used to effectively manage image contents, and image search can be as efficient as text search by transforming images into textual documents through machine learning. Experiments in both the Corel dataset and real Web dataset are performed to validate our system and the results are promising. This system suggests a new combination of texts and visual features in order to achieve a semantic image search, and is expected to become a re-ranking system to the existing image search result via the Internet.