Multi-sensory face biometric fusion (for personal identification)

The objective of this work is to recognize faces using sets of images in visual and thermal spectra. This is challenging because the former is greatly affected by illumination changes, while the latter frequently contains occlusions due to eye-wear and is inherently less discriminative. Our method is based on a fusion of the two modalities. Specifically: we examine (i) the effects of preprocessing of data in each domain, (ii) the fusion of holistic and local facial appearance, and (iii) propose an algorithm for combining the similarity scores in visual and thermal spectra in the presence of prescription glasses and significant pose variations, using a small number of training images (5-7). Our system achieved a high correct identification rate of 97% on a freely available test set of 29 individuals and extreme illumination changes.

An adaptive approach to information discovery in multi-dimensional wireless sensor networks

Multidimensional WSNs are deployed in complex environments to sense and collect data relating to multiple attributes (multidimensional data). Such networks present unique challenges to data dissemination, data storage and in-network query processing (information discovery). In this paper, we investigate efficient strategies for information discovery in large-scale multidimensional WSNs and propose the Adaptive MultiDimensional Multi-Resolution Architecture (A-MDMRA) that efficiently combines “push” and “pull” strategies for information discovery and adapts to variations in the frequencies of events and queries in the network to construct optimal routing structures. We present simulation results showing the optimal routing structure depends on the frequency of events and query occurrence in the network. It also balances push and pull operations in large scale networks enabling significant QoS improvements and energy savings.

Multi-phase microstructure design of a novel high strength TRIP steel through experimental methodology

The multi-phase structure of a novel low-alloy transformation induced plasticity (TRIP) steel was designed through experimental analysis. The evolutions of both microstructure and mechanical properties during the two-stage heat treatment were analyzed. The phase transformations during the intercritical annealing and the isothermal bainitic transformation were investigated by means of dilatometry. It was shown that two types of C diffusion were detected during intercritical annealing and a complex microstructure was formed after heat treatment. The processing parameters were selected in such a way to obtain microstructures with systematically different volume fractions of ferrite, bainite and retained austenite. The volume fractions of ferrite and retained austenite were found to be two main factors controlling the ductility. Furthermore, a high volume fraction of C-rich retained austenite, which was stabilized at room temperature, was the origin of a TRIP effect. The resulting material demonstrates a significant improvement in the ultimate tensile strength (1077. MPa) with good uniform elongation (22.5%), as compared to conventional TRIP steels. © 2014 Elsevier B.V.

Energy efficient information discovery approach for range queries in multi-dimensional WSNs

Multidimensional WSNs are deployed in complex environments to sense and collect data relating to multiple attributes (multi-dimensional data). Such networks present unique challenges to data dissemination, data storage and in-network query processing (information discovery). Recent algorithms proposed for such WSNs are aimed at achieving better energy efficiency and minimizing latency. This creates a partitioned network area due to the overuse of certain nodes in areas which are on the shortest or closest or path to the base station or data aggregation points which results in hotspots nodes. In this paper, we propose a time-based multi-dimensional, multi-resolution storage approach for range queries that balances the energy consumption by balancing the traffic load as uniformly as possible. Thus ensuring a maximum network lifetime. We present simulation results to show that the proposed approach to information discovery offers significant improvements on information discovery latency compared with current approaches. In addition, the results prove that the Quality of Service (QoS) improvements reduces hotspots thus resulting in significant network-wide energy saving and an increased network lifetime.

Information discovery in multi-dimensional autonomous wireless sensor networks

 The thesis proposed four novel algorithms of information discovery for Multidimensional Autonomous Wireless Sensor Networks (WSNs) that can significantly increase network lifetime and minimize query processing latency, resulting in quality of service improvements that are of immense benefit to Multidimensional Autonomous WSNs are deployed in complex environments (e.g., mission-critical applications).

Multi-output interval type-2 fuzzy logic system for protein secondary structure prediction

A new multi-output interval type-2 fuzzy logic system (MOIT2FLS) is introduced for protein secondary structure prediction in this paper. Three outputs of the MOIT2FLS correspond to three structure classes including helix, strand (sheet) and coil. Quantitative properties of amino acids are employed to characterize twenty amino acids rather than the widely used computationally expensive binary encoding scheme. Three clustering tasks are performed using the adaptive vector quantization method to construct an equal number of initial rules for each type of secondary structure. Genetic algorithm is applied to optimally adjust parameters of the MOIT2FLS. The genetic fitness function is designed based on the Q3 measure. Experimental results demonstrate the dominance of the proposed approach against the traditional methods that are Chou-Fasman method, Garnier-Osguthorpe-Robson method, and artificial neural network models.

Multi-dimensional liquid chromatography and metabolomics, are two dimensions better than one?

 While data processing methods in metabolomic studies often work with 'n' number of dimensions, analytical techniques, with the notable exception of NMR, have mostly stuck only to one. Peak overlap continues to be a problem and there is an ever-present demand to maximize the number of metabolites that can be separated and identified in a single run. One method that might help to overcome these issues is multidimensional liquid chromatography, which uses two columns of different phases. A sequential collection of aliquots is made from the first column and reinjected onto a second, and the resulting data are then plotted in 2D or 3D space. The total peak capacity of such a system is the combined peak capacities of each column. The 'offline' version of this technique, using a fraction collector, was introduced over 30 years ago but with recent advances in instrumentation and software, particularly the 'online' approach using automated switching valves, has led to increasing interest in the technique. Both offline and online methods can be carried out as a comprehensive procedure, or via 'heart-cutting', in which only specific peaks are analysed in the second dimension. Past applications include proteomics, natural product chemistry, forensic science and pharmaceutical analysis. These successes are likely to be built on in the future as new column chemistries and bio-informatic approaches are developed. In this review an overview of the theory of twodimensional liquid chromatography is presented, its potential in the field of metabolomics is assessed and predictions for future research directions are made.

Emotion processing fails to modulate putative mirror neuron response to trained visuomotor associations

Recent neuroimaging studies have demonstrated that activation of the putative human mirror neuron system (MNS) can be elicited via visuomotor training. This is generally interpreted as supporting an associative learning account of the mirror neuron system (MNS) that argues against the ontogeny of the MNS to be an evolutionary adaptation for social cognition. The current study assessed whether a central component of social cognition, emotion processing, would influence the MNS activity to trained visuomotor associations, which could support a broader role of the MNS in social cognition. Using functional magnetic resonance imaging (fMRI), we assessed repetition suppression to the presentation of stimulus pairs involving a simple hand action and a geometric shape that was either congruent or incongruent with earlier association training. Each pair was preceded by an image of positive, negative, or neutral emotionality. In support of an associative learning account of the MNS, repetition suppression was greater for trained pairs compared with untrained pairs in several regions, primarily supplementary motor area (SMA) and right inferior frontal gyrus (rIFG). This response, however, was not modulated by the valence of the emotional images. These findings argue against a fundamental role of emotion processing in the mirror neuron response, and are inconsistent with theoretical accounts linking mirror neurons to social cognition.

Mixed-norm sparse representation for multi view face recognition

Face recognition with multiple views is a challenging research problem. Most of the existing works have focused on extracting shared information among multiple views to improve recognition. However, when the pose variation is too large or missing, 'shared information' may not be properly extracted, leading to poor recognition results. In this paper, we propose a novel method for face recognition with multiple view images to overcome the large pose variation and missing pose issue. By introducing a novel mixed norm, the proposed method automatically selects candidates from the gallery to best represent a group of highly correlated face images in a query set to improve classification accuracy. This mixed norm combines the advantages of both sparse representation based classification (SRC) and joint sparse representation based classification (JSRC). A trade off between the ℓ1-norm from SRC and ℓ2,1-norm from JSRC is introduced to achieve this goal. Due to this property, the proposed method decreases the influence when a face image is unseen and has large pose variation in the recognition process. And when some face images with a certain degree of unseen pose variation appear, this mixed norm will find an optimal representation for these query images based on the shared information induced from multiple views. Moreover, we also address an open problem in robust sparse representation and classification which is using ℓ1-norm on the loss function to achieve a robust solution. To solve this formulation, we derive a simple, yet provably convergent algorithm based on the powerful alternative directions method of multipliers (ADMM) framework. We provide extensive comparisons which demonstrate that our method outperforms other state-of-the-arts algorithms on CMU-PIE, Yale B and Multi-PIE databases for multi-view face recognition.