Earticle

The average mutual information (AMI) known from information theory has been reported as a strong genome signature in some literature and we have reported the use of oligonucleotide frequencies as a genome signature. In this work we improve the use of AMI as a training feature for Growing Self Organising Maps (GSOM). Although the range of k is considered as an important parameter in AMI, no standard range for k is proposed. Our first contribution is to introduce a new growth threshold (GT) for GSOM and use it to identify the best range of k for clustering prokaryotic sequence fragments of 10 kb. We then, compare the results using the best k range of AMI against our previously published results using oligonucleotide frequencies. These experiments showed that the newly proposed GT equation makes GSOM able to efficiently and effectively analyse different data features for the same data. The results also emphasize our use of oligonucleotide frequencies as opposed to AMI.

This paper deals with a new approach for detection of finger veins. The method is split into four subsequent steps. The first of them consists of basic series of image filtering of the vein pattern and the other three are sequences of image filters for determination of the finger contour used for the background masking. At the end the finger veins with finger contour are extracted and sent to the template generation unit.

DNA microarray is a multiplex technology used in molecular biology and biomedicine. It consists of an arrayed series of thousands of microscopic spots of DNA oligonucleotides, called features, of which the result should be analyzed by computational methods. Analyzing microarray data using intelligent computing methods has attracted many researchers in recent years. Several approaches have been proposed, in which machine learning based approaches play an important role for biomedical research such as gene expression interpretation, classification and prediction for cancer diagnosis, etc. In this paper, we present an application of the feedforward neural network (SLFN) trained by the singular value decomposition (SVD) approach for DNA microarray classification. The classifier of the single hidden-layer feedforward neural network (SLFN) has the activation function of the hidden units to be ‘tansig’. Experimental results show that the SVD trained feedforward neural network is simple in both training procedure and network structure; it has low computational complexity and can produce better performance with compact network architecture.

This paper focuses on the skin physiological factor relationship analysis using grey GM(1, N) and GM(0, N) model. First, skin physiological factor sampling are done for Taiwan female subjects aged from 18 ~ 52, where the factors include skin elasticity, pH value, skin pigmentation, skin surface lipids and skin epidermal hydration. With the acquired data, we can then establish the data model using grey theory. Here, we apply the grey GM(1, N) and GM(0, N) model to obtain the relationship weighting between the major factor and the other relational factors. Furthermore, according to the determined weightings, we proceed on the skin physiological factor relationship analysis to understand the skin characteristics under different age.

To understand the structure-to-function relationship, life sciences researchers and biologists need to retrieve similar structures from protein databases and classify them into the same protein fold. With the technology innovation the number of protein structures increases every day, so, retrieving structurally similar proteins using current structural alignment algorithms may take hours or even days. Therefore, improving the efficiency of protein structure retrieval and classification becomes an important research issue. In this paper we propose novel approach which provides faster classification (minutes) of protein structures. We build separate Hidden Markov Model (HMM) for each class. In our approach we align tertiary structures of proteins. Viterbi algorithm is used to find the most probable path to the model. We have compared our approach against an existing approach named 3D HMM, which also performs alignment of tertiary structures of proteins by using HMM. The results show that our approach is more accurate than 3D HMM.

Conventional multimodal biometric identification systems tend to have larger memory footprint, slower processing speeds and a higher implementation and operational cost. In this paper we propose a state of the art framework for multimodal biometric identification system which can be adapted for any type of biometrics to provide smaller memory footprint and faster implementation than the conventional multimodal biometrics systems. The proposed framework is verified by development of a fingerprint and iris based fusion system which utilizes a single Hamming Distance matcher. Extensive testing is performed on the system running in identification mode and the results show that the system not only provides higher accuracy than the individual unimodal system but also the results are comparable to the conventional system.

Bringing a criminal to justice is a labour intensive process. In the current paper, we explored ways of reducing the police time involved with the construction and identification of facial composites (these are images of wanted persons made by witnesses and victims of crime). A software system called EvoFIT was used that ‘evolved’ a composite by the repeated selection and breeding of complete faces. In the first part, a standalone version of EvoFIT was designed and evaluated in the laboratory. This performed similarly to the full system that normally requires several hours of a police officer’s time. It was also found that composite quality did not change overall, although was more variable in one of the measures used, if users were asked to make fast rather than slow face selection judgements. In latter work, a small database of composites was built that could be used to search for matching identities. It was found that pixel intensity (texture) information was valuable for composites produced from a traditional ‘feature’ based system, but feature shape information was valuable for composites produced from EvoFIT. The results show promise for the automated construction and identification of composite images.