Fisher Discriminant Algorithm Research Articles

Extraction of Optimal Measurements for Drowsy Driving Detection considering Driver Fingerprinting Differences

Contributions of measurements for detecting drowsy driving are determined by calculation parameters, which are directly related to the accuracy of drowsiness detection. The previous studies utilized the same Unified Calculation Parameters (UCPs) to compute each driver’s measurements. However, since each driver has unique driving behavior characteristics, namely, driver fingerprinting, Individual Drivers’ Best Calculation Parameters (IDBCPs) making measurements more discriminative for drowsiness are various. Regardless of the difference in driver fingerprinting among the drivers being tested, using UCPs instead of IDBCPs to compute measurements will limit the drowsiness-detection performance of the measurements and reduce drowsiness-detection accuracies at the individual driver level. Thus, this paper proposed a model to optimize calculation parameters of individual driver’s measurements and to extract individual driver’s measurements that effectively distinguish drowsy driving. Through real vehicle experiments, we collected naturalistic driving data and subjective drowsy levels evaluated by the Karolinska Sleepiness Scale. Eight nonintrusive drowsiness-related measurements were calculated by double-layer sliding time windows. In the proposed model, we firstly applied the Wilcoxon test to analyze differences between measurements of the awake state and drowsy state, and constructed the fitness function reflecting the relationship between the calculation parameters and measurement’s drowsiness-detection performance. Secondly, the genetic algorithms were used to optimize fitness functions to obtain measured IDBCPs. Finally, we selected measurements calculated by IDBCPs that can distinguish drowsy driving to constitute individual drivers’ optimal drowsiness-detection measurement set. To verify the advantages of IDBCPs, the measurements calculated by UCPs and IDBCPs were, respectively, used to build driver-specific drowsiness-detection models: DF_U and DF_I based on the Fisher discriminant algorithm. The mean drowsiness-detection accuracies of DF_U and DF_I were, respectively, 85.25% and 91.06%. It indicated that IDBCPs could enhance measurements’ drowsiness-detection performance and improve the drowsiness-detection accuracies. This paper contributed to the establishment of personalized drowsiness-detection models considering driver fingerprinting differences.

Polychotomous kernel Fisher discriminant via top–down induction of binary tree

In spite of the popularity of Fisher discriminant analysis in the realm of feature extraction and pattern classification, it is beyond the capability of Fisher discriminant analysis to extract nonlinear structures from the data. That is where the kernel Fisher discriminant algorithm sets in the scenario of supervised learning. In this article, a new trail is blazed in developing innovative and effective algorithm for polychotomous kernel Fisher discriminant with the capability in estimating the posterior probabilities, which is exceedingly necessary and significant in solving complex nonlinear pattern recognition problems arising from the real world. Different from the conventional ‘divide-and-combine’ approaches to polychotomous classification problems, such as pairwise and one-versus-others, the method proposed herein synthesizes the multi-category classifier via the induction of top-to-down binary tree by means of kernelized group clustering algorithm. The deficiencies inherited in the conventional multi-category kernel Fisher discriminant are surmounted and the simulation on a benchmark image dataset demonstrates the superiority of the proposed approach.

Prediction of protein structural classes for low-homology sequences based on predicted secondary structure

BackgroundPrediction of protein structural classes (α, β, α + β and α/β) from amino acid sequences is of great importance, as it is beneficial to study protein function, regulation and interactions. Many methods have been developed for high-homology protein sequences, and the prediction accuracies can achieve up to 90%. However, for low-homology sequences whose average pairwise sequence identity lies between 20% and 40%, they perform relatively poorly, yielding the prediction accuracy often below 60%.ResultsWe propose a new method to predict protein structural classes on the basis of features extracted from the predicted secondary structures of proteins rather than directly from their amino acid sequences. It first uses PSIPRED to predict the secondary structure for each protein sequence. Then, the chaos game representation is employed to represent the predicted secondary structure as two time series, from which we generate a comprehensive set of 24 features using recurrence quantification analysis, K-string based information entropy and segment-based analysis. The resulting feature vectors are finally fed into a simple yet powerful Fisher's discriminant algorithm for the prediction of protein structural classes. We tested the proposed method on three benchmark datasets in low homology and achieved the overall prediction accuracies of 82.9%, 83.1% and 81.3%, respectively. Comparisons with ten existing methods showed that our method consistently performs better for all the tested datasets and the overall accuracy improvements range from 2.3% to 27.5%. A web server that implements the proposed method is freely available at http://www1.spms.ntu.edu.sg/~chenxin/RKS_PPSC/.ConclusionThe high prediction accuracy achieved by our proposed method is attributed to the design of a comprehensive feature set on the predicted secondary structure sequences, which is capable of characterizing the sequence order information, local interactions of the secondary structural elements, and spacial arrangements of α helices and β strands. Thus, it is a valuable method to predict protein structural classes particularly for low-homology amino acid sequences.

Cluster protein structures using recurrence quantification analysis on coordinates of alpha-carbon atoms of proteins

The 3-dimensional coordinates of alpha-carbon atoms of proteins are used to distinguish the protein structural classes based on recurrence quantification analysis (RQA). We consider two independent variables from RQA of coordinates of alpha-carbon atoms, %determ1 and %determ2, which were defined by Webber et al. [C.L. Webber Jr., A. Giuliani, J.P. Zbilut, A. Colosimo, Proteins Struct. Funct. Genet. 44 (2001) 292]. The variable %determ2 is used to define two new variables, %determ21 and %determ22. Then three variables %determ1, %determ21 and %determ22 are used to construct a 3-dimensional variable space. Each protein is represented by a point in this variable space. The points corresponding to proteins from the α, β, α+β and α/β structural classes position into different areas in this variable space. In order to give a quantitative assessment of our clustering on the selected proteins, Fisher's discriminant algorithm is used. Numerical results indicate that the discriminant accuracies are very high and satisfactory.

A reformative kernel Fisher discriminant algorithm and its application to face recognition

In this paper, a reformative kernel Fisher discriminant (KFD) algorithm with fuzzy set theory is studied. The KFD algorithm is effective to extract nonlinear discriminative features of input samples using the kernel trick. However, the conventional KFD algorithm assumes the same level of relevance of each sample to the corresponding class. In this paper, a fuzzy kernel Fisher discriminant (FKFD) algorithm is proposed. Distribution information of samples is represented with fuzzy membership degree and this information is utilized to redefine corresponding scatter matrices which are different to the conventional KFD algorithm and effective to extract discriminative features from overlapping (outlier) samples. Experimental results on the ORL face database demonstrate the effectiveness of the proposed method.

A fractal method to distinguish coding and non-coding sequences in a complete genome based on a number sequence representation

A fractal method to distinguish coding and non-coding sequences in a complete genome is proposed, based on different statistical behaviors between these two kinds of sequences. We first propose a number sequence representation of DNA sequences. Multifractal analysis is then performed on the measure representation of the obtained number sequence. The three exponents C - 1 , C 1 and C 2 are selected from the result of multifractal analysis. Each DNA may be represented by a point in the three-dimensional space generated by these three-component vectors. It is shown that points corresponding to coding and non-coding sequences in the complete genome of many prokaryotes are roughly distributed in different regions. Fisher's discriminant algorithm can be used to separate these two regions in the spanned space. If the point ( C - 1 , C 1 , C 2 ) for a DNA sequence is situated in the region corresponding to coding sequences, the sequence is discriminated as a coding sequence; otherwise, the sequence is classified as a non-coding one. For all 51 prokaryotes we considered , the average discriminant accuracies p c , p nc , q c and q nc reach 72.28%, 84.65%, 72.53% and 84.18%, respectively.

A new kernel Fisher discriminant algorithm with application to face recognition

Kernel-based methods have been of wide concern in the field of machine learning and neurocomputing. In this paper, a new Kernel Fisher discriminant analysis (KFD) algorithm, called complete KFD (CKFD), is developed. CKFD has two advantages over the existing KFD algorithms. First, its implementation is divided into two phases, i.e., Kernel principal component analysis (KPCA) plus Fisher linear discriminant analysis (FLD), which makes it more transparent and simpler. Second, CKFD can make use of two categories of discriminant information, which makes it more powerful. The proposed algorithm was applied to face recognition and tested on a subset of the FERET database. The experimental results demonstrate that CKFD is significantly better than the algorithms of Kernel Fisherface and Kernel Eigenface.

A new quantitative criterion to distinguish between α/β and α+β proteins (domains)

According to the statistical analysis, it is shown that the differences of the content of α-helix and β-strand between α/β and α+β proteins are of statistical significance. Based on the secondary structure content and the percentage of parallel or anti-parallel strands, any mixed αβ protein can be represented by a point in a three-dimensional prism. The distribution of the mapping points for 79 mixed αβ proteins (domains), of which 26 are class α/β and 53 are class α+β, shows that the two kinds of points are situated at distinct regions roughly. A new quantitative criterion based on the Fisher discriminant algorithm is proposed to distinguish between the α/β and α+β proteins (domains). Of the 79 proteins 77 are correctly classified (97.5%). As a stringent cross-validation test, the jackknife test shows that of the 79 proteins 77 are correctly classified. The jackknife test accuracy is still 97.5%. These figures indicate the self-consistence and the extrapolating effectiveness of the new quantitative criterion. Applying the new criterion to reclassify the α/β and α+β proteins (domains) in SCOP is also discussed. It is hoped that the new quantitative criterion will be useful for the development of protein classification databases.

A Novel Approach to Distinguish Between Intron-containing and Intronless Genes Based on the Format of Z Curves

A novel method to distinguish between intron-containing and intronless DNA sequences has been proposed, based on different statistic behaviors between them. In this method, DNA sequences are first represented as Z curves. Three exponents α, β and γ for each given sequence are calculated based on the format of the Z curve for the DNA sequence. A three-dimensional space is spanned by the three exponents. Each DNA sequence may be represented by a point in this space. One hundred intronless and intron-containing genes, respectively, were selected randomly from the GenBank or EMBL database. It is shown that the 200 points are roughly distributed in different regions. The best separating plane to separate the two regions is obtained by using Fisher's discriminant algorithm. For any given sequence to be discriminated, calculate three exponents α, β and γ, corresponding to a point in the three-dimensional space. It the point is situated at the upper region of the separating plane, the sequence is discriminated as an intronless one; otherwise, the sequence is an intron-containing one. A test of the method for the sequences in an independent test set shows that the discriminant accuracy reaches as high as 89.0%.