International Journal of Hybrid Intelligent Systems

Why do people feel happy and good or empathize more, with smiling faces than with expressionless faces? To understand the mechanism involved in the recognition of facial imagery, we evaluated the degree of favorability for 46 expressionless and smiling facial images obtained from young persons including 84 subjects having no pre-acquired knowledge to the experiments. Each image was presented to every subject and was asked to rank favorability on a scale from 1 to 10. By using of two types of fractal analysis, i.e., planner and cubic analyses, fractal dimensions of facial images were obtained in order to characterize the complexity of the images. Furthermore, we calculated the local fractal dimensions around parts of eyes and mouths. The results showed a significant difference in the local fractal dimension around mouths between the expressionless and smiling faces. On the other hand, we found a strong correlation between the degree of favorability and fractal dimensions of whole faces. However, a significant correlation between the local fractal dimensions around mouths and favorability was not found. These showed that humans recognized face in the wholly way, rather than locally, when we defined the favorability of facial imagery. These results imply that the fractal dimension obtained in relation to complexity in imagery optically information is useful in characterizing the psychological processes of cognition and awareness.

One approach to investigating neural death is through systematic studies of the changing morphology of cultured brain neurons in response to cellular challenges. Image segmentation and neuron skeleton reconstruction methods developed to date to analyze such changes have been limited by the low contrast of cells. In this paper we present new algorithms that successfully circumvent these problems. The binary method is based on logical analysis of grey and distance difference of images. The spurious regions are detected and removed through use of a hierarchical window filter. The skeletons of binary cell images are extracted. The extension direction and connection points of broken cell skeletons are automatically determined, and broken neural skeletons are reconstructed. The spurious strokes are deleted based on cell prior knowledge. The reconstructed skeletons are processed furthermore by filling holes, smoothing and extracting new skeletons. The final constructed neuron skeletons are analyzed and calculated to find the length and morphology of skeleton branches automatically. The efficacy of the developed algorithms is demonstrated here through a test of cultured brain neurons from newborn mice.

In mass spectrometry (MS) analysis, false peak detection results are unavoidable due to severe spectrum variations. However, most current peak detection methods are neither robust enough to resist spectrum variations nor flexible enough to revise false detection results. To solve these two problems, we first propose peak tree to reveal the hierarchical relation among peak judgments made on different scales. Different peak tree decomposition will lead to different peak detection result, which make it very convenient to revise false result. Then, we propose a closed-loop scheme to iteratively refine peak tree decomposition. Experiment results show that, compared with conventional peak detection methods, our method can better resist spectrum variations and provide a more consistent result among different spectra.

Cerebral palsy (CP) is generally considered as a non-progressive neuro-developmental condition that occurs in early childhood and is associated with a motor impairment, usually affecting mobility and posture. Automatic accurate identification of cerebral palsy gait has many potential applications, for example, assistance in diagnosis, clinical decision-making and communication among the clinical professionals. In previous studies, support vector machine (SVM) and some other pattern classification methods like neural networks have been applied to classify CP gait patterns. The objective of this study is to first further investigate different classification paradigms in the CP gait analysis, particularly the Kernel Fisher Discriminant Analysis (KFD) which has been successfully applied to many pattern recognition problems and identified as a strong competitor of SVM. The component obtained by KFD maximally separates two classes in the feature space, thus overcoming the limitations of linear discriminant analysis of being unable to extract nonlinear features representing higher-order statistics. Using a publicly available CP gait dataset (68 normal healthy and 88 with spastic diplegia form of CP), a comprehensive performances comparison was presented with different features including the two basic temporal-spatial gait parameters (stride length and cadence). Various cross-validation testing show that the KFD offers better classification accuracies than the support vector machine and is superior to a number of other classification methods such as decision tree, multiple layer perceptron and k nearest neighbor.

Rough set theory (RST) is concerned with the formal approximation of crisp sets and is a mathematical tool which deals with vagueness and uncertainty. This paper presents an approach to optimize rough set partition sizes using various optimization techniques. The forecasting accuracy is measured by using the area under the curve (AUC) of the receiver operating characteristic (ROC) curve. The four optimization techniques used are genetic algorithm, particle swarm optimization, hill climbing and simulated annealing. This proposed method is tested on two data sets, namely, the human immunodeficiency virus (HIV) data set and the militarized interstate dispute (MID) data set. The results obtained from this granulization method are compared to two previous static granulization methods, namely, equal-width-bin and equal-frequency-bin partitioning. The results conclude that all of the proposed optimized methods produce higher forecasting accuracies than that of the two static methods. In the case of the HIV data set, the hill climbing approach produced the highest accuracy; an accuracy of 69.02% is achieved in a time of 210.4 hours. For the MID data, the genetic algorithm approach produced the highest accuracy. The accuracy achieved is 95.82% in a time of 7 hours. The rules generated from the rough set are linguistic and easy-to-interpret, but this does come at the expense of the accuracy lost in the discretization process where the granularity of the variables is decreased.