a b s t r a c tFeature selection is used in many application areas relevant to expert and intelligent systems, such as data mining and machine learning, image processing, anomaly detection, bioinformatics and natural language processing. Feature selection based on information theory is a popular approach due its computational efficiency, scalability in terms of the dataset dimensionality, and independence from the classifier. Common drawbacks of this approach are the lack of information about the interaction between the features and the classifier, and the selection of redundant and irrelevant features. The latter is due to the limitations of the employed goal functions leading to overestimation of the feature significance.To address this problem, this article introduces two new nonlinear feature selection methods, namely Joint Mutual Information Maximisation (JMIM) and Normalised Joint Mutual Information Maximisation (NJMIM); both these methods use mutual information and the 'maximum of the minimum' criterion, which alleviates the problem of overestimation of the feature significance as demonstrated both theoretically and experimentally. The proposed methods are compared using eleven publically available datasets with five competing methods. The results demonstrate that the JMIM method outperforms the other methods on most tested public datasets, reducing the relative average classification error by almost 6% in comparison to the next best performing method. The statistical significance of the results is confirmed by the ANOVA test. Moreover, this method produces the best trade-off between accuracy and stability.

Quantitative assessment of movement impairment in Huntington’s disease (HD) is essential to monitoring of disease progression. This paper aimed to develop and validate a novel low cost, objective automated system for the evaluation of upper limb movement impairment in HD in order to eliminate the inconsistency of the assessor and offer a more sensitive, continuous assessment scale. Patients with genetically confirmed HD and healthy controls were recruited to this observational study. Demographic data, including age (years), gender, and unified HD rating scale total motor score (UHDRS-TMS), were recorded. For the purposes of this paper, a modified upper limb motor impairment score (mULMS) was generated from the UHDRS-TMS. All participants completed a brief, standardized clinical assessment of upper limb dexterity while wearing a tri-axial accelerometer on each wrist and on the sternum. The captured acceleration data were used to develop an automatic classification system for discriminating between healthy and HD participants and to automatically generate a continuous movement impairment score (MIS) that reflected the degree of the movement impairment. Data from 48 healthy and 44 HD participants was used to validate the developed system, which achieved 98.78% accuracy in discriminating between healthy and HD participants. The Pearson correlation coefficient between the automatic MIS and the clinician rated mULMS was 0.77 with a p-value < 0.01. The approach presented in this paper demonstrates the possibility of an automated objective, consistent, and sensitive assessment of the HD movement impairment.

The introduction of heterogeneous wireless mesh technologies provides an opportunity for higher network capacity, wider coverage, and higher quality of service (QoS). Each wireless device utilizes different standards, data formats, protocols, and access technologies. However, the diversity and complexity of such technologies create challenges for traditional control and management systems. This paper proposes a heterogeneous metropolitan area network architecture that combines an IEEE 802.11 wireless mesh network (WMN) with a long-term evolution (LTE) network. In addition, a new heterogeneous routing protocol and a routing algorithm based on reinforcement learning called cognitive heterogeneous routing are proposed to select the appropriate transmission technology based on parameters from each network. The proposed heterogeneous network overcomes the problems of sending packets over long paths, island nodes, and interference in WMNs and increases the overall capacity of the combined network by utilizing unlicensed frequency bands instead of buying more license frequency bands for LTE. The work is validated through extensive simulations that indicate that the proposed heterogeneous WMN outperforms the LTE and Wi-Fi networks when used individually. The simulation results show that the proposed network achieves an increase of up to 200% in throughput compared with Wi-Fi-only networks or LTE-only networks.

No abstract

We

For asymptomatic patients suffering from carotid stenosis, the assessment of plaque morphology is an important clinical task which allows monitoring of the risk of plaque rupture and future incidents of stroke. Ultrasound Imaging provides a safe and non-invasive modality for this, and the segmentation of media-adventitia boundaries and lumen-intima boundaries of the Carotid artery form an essential part in this monitoring process. In this paper, we propose a novel Deep Convolutional Neural Network as a fully automated segmentation tool, and its application in delineating both the media-adventitia boundary and the lumen-intima boundary. We develop a new geometrically constrained objective function as part of the Network's Stochastic Gradient Descent optimisation, thus tuning it to the problem at hand. Furthermore, we also apply a novel, bimodal fusion of envelope and phase congruency data as an input to the network, as the latter provides an intensity-invariant data source to the network. We finally report the segmentation performance of the network on transverse sections of the carotid. Tests are carried out on an augmented dataset of 81,000 images, and the results are compared to other studies by reporting the DICE coefficient of similarity, modified Hausdorff Distance, sensitivity and specificity. Our proposed method is shown to yield results of comparable accuracy over this larger dataset, with the advantage of it being fully automated. We conclude that Deep Convolutional Neural Networks provide a reliable trained manner in which carotid ultrasound images may be automatically segmented, using envelope data and intensity invariant phase congruency maps as a data source.

No abstract

No abstract