Youxi Wu

Classification of Mental Task From EEG Signals Using Immune Feature Weighted Support Vector Machines

The classification of mental tasks is one of key issues of EEG-based brain computer interface (BCI). Differentiating classes of mental tasks from EEG signals is challenging because EEG signals are nonstationary and nonlinear. Owing to its powerful capacity in solving nonlinearity problems, support vector machine (SVM) method has been widely used as a classification tool. Traditional SVMs, however, assume that each feature of a sample contributes equally to classification accuracy, which is not necessarily true in real applications. In addition, the parameters of SVM and the kernel function also affect classification accuracy. In this study, immune feature weighted SVM (IFWSVM) method was proposed. Immune algorithm (IA) was then introduced in searching for the optimal feature weights and the parameters simultaneously. IFWSVM was used to multiclassify five different mental tasks. Theoretical analysis and experimental results showed that IFWSVM has better performance than traditional SVM.

Subkilometer crater discovery with boosting and transfer learning

Counting craters in remotely sensed images is the only tool that provides relative dating of remote planetary surfaces. Surveying craters requires counting a large amount of small subkilometer craters, which calls for highly efficient automatic crater detection. In this article, we present an integrated framework on autodetection of subkilometer craters with boosting and transfer learning. The framework contains three key components. First, we utilize mathematical morphology to efficiently identify crater candidates, the regions of an image that can potentially contain craters. Only those regions occupying relatively small portions of the original image are the subjects of further processing. Second, we extract and select image texture features, in combination with supervised boosting ensemble learning algorithms, to accurately classify crater candidates into craters and noncraters. Third, we integrate transfer learning into boosting, to enhance detection performance in the regions where surface morphology differs from what is characterized by the training set. Our framework is evaluated on a large test image of 37,500 × 56,250 m2 on Mars, which exhibits a heavily cratered Martian terrain characterized by nonuniform surface morphology. Empirical studies demonstrate that the proposed crater detection framework can achieve an F1 score above 0.85, a significant improvement over the other crater detection algorithms.

Mining sequential patterns with periodic wildcard gaps

Mining frequent patterns with periodic wildcard gaps is a critical data mining problem to deal with complex real-world problems. This problem can be described as follows: given a subject sequence, a pre-specified threshold, and a variable gap-length with wildcards between each two consecutive letters. The task is to gain all frequent patterns with periodic wildcard gaps. State-of-the-art mining algorithms which use matrices or other linear data structures to solve the problem not only consume a large amount of memory but also run slowly. In this study, we use an Incomplete Nettree structure (the last layer of a Nettree which is an extension of a tree) of a sub-pattern P to efficiently create Incomplete Nettrees of all its super-patterns with prefix pattern P and compute the numbers of their supports in a one-way scan. We propose two new algorithms, MAPB (Mining sequentiAl Pattern using incomplete Nettree with Breadth first search) and MAPD (Mining sequentiAl Pattern using incomplete Nettree with Depth first search), to solve the problem effectively with low memory requirements. Furthermore, we design a heuristic algorithm MAPBOK (MAPB for tOp-K) based on MAPB to deal with the Top-K frequent patterns for each length. Experimental results on real-world biological data demonstrate the superiority of the proposed algorithms in running time and space consumption and also show that the pattern matching approach can be employed to mine special frequent patterns effectively.

Tumor Detection in MR Images Using One-Class Immune Feature Weighted SVMs

Tumor detection using medical images plays a key role in medical practices. One challenge in tumor detection is how to handle the nonlinear distribution of the real data. Owing to its ability of learning the nonlinear distribution of the tumor data without using any prior knowledge, one-class support vector machines (SVMs) have been applied in tumor detection. The conventional one-class SVMs, however, assume that each feature of a sample has the same importance degree for the classification result, which is not necessarily true in real applications. In addition, the parameters of one-class SVM and its kernel function also affect the classification result. In this study, immune algorithm (IA) was introduced in searching for the optimal feature weights and the parameters simultaneously. One-class immune feature weighted SVM (IFWSVM) was proposed to detect tumors in MR images. Theoretical analysis and experimental results showed that one-class IFWSVM has better performance than conventional one-class SVM.

Automatic detection of craters in planetary images: an embedded framework using feature selection and boosting

Identifying impact craters on planetary surfaces is one fundamental task in planetary science. In this paper, we present an embedded framework on auto-detection of craters, using feature selection and boosting strategies. The paradigm aims at building a universal and practical crater detector. This methodology addresses three issues that such a tool must possess: (i) it utilizes mathematical morphology to efficiently identify the regions of an image that can potentially contain craters; only those regions, defined as crater candidates, are the subjects of further processing; (ii) it selects Haar-like image texture features in combination with boosting ensemble supervised learning algorithms to accurately classify candidates into craters and non-craters; (iii) it uses transfer learning, at a minimum additional cost, to enable maintaining an accurate auto-detection of craters on new images, having morphology different from what has been captured by the original training set. All three aforementioned components of the detection methodology are discussed, and the entire framework is evaluated on a large test image of 37,500 x 56,250

FP-ELM

m2 on Mars, showing heavily cratered Martian terrain characterized by nonuniform surface morphology. Our study demonstrates that this methodology provides a robust and practical tool for planetary science, in terms of both detection accuracy and efficiency.

A Nettree for pattern Matching with flexible wildcard Constraints

The online sequential extreme learning machine (OS-ELM) algorithm is an on-line and incremental learning method, which can learn data one-by-one or chunk-by-chunk with a fixed or varying chunk size. And OS-ELM achieves the same learning performance as ELM trained by the complete set of data. However, in on-line learning environments, the concepts to be learned may change with time, a feature referred to as concept drift. To use ELMs in such non-stationary environments, a forgetting parameters extreme learning machine (FP-ELM) is proposed in this paper. The proposed FP-ELM can achieve incremental and on-line learning, just like OS-ELM. Furthermore, FP-ELM will assign a forgetting parameter to the previous training data according to the current performance to adapt to possible changes after a new chunk comes. The regularized optimization method is used to avoid estimator windup. Performance comparisons between FP-ELM and two frequently used ensemble approaches are carried out on several regression and classification problems with concept drift. The experimental results show that FP-ELM produces comparable or better performance with lower training time.

Research on the Segmentation of MRI Image Based on Multi-Classification Support Vector Machine

In this paper, a new nonlinear structure called Nettree is proposed. A Nettree is different from a tree in that a node may have more than one parent. An algorithm, named Nettree for pattern Matching with flexible wildcard Constraints (NAMEIC), based on Nettree is designed to solve pattern matching with flexible wildcard constraints. The problem is exponential with regard to the pattern length m. We prove the correctness of the algorithm, and illustrate how it works through an example. NAMEIC is W*m times faster than an existing approach because the result can be given after creating the Nettree in one pass, where W is the maximal gap flexibility. Experiments validate the correctness and efficiency of NAMEIC.

The Apriori property of sequence pattern mining with wildcard gaps

In head MRI image, the boundary of each encephalic tissue is highly complicated and irregular. It is a real challenge to traditional segmentation algorithms. As a new kind of machine learning, support vector machine (SVM) based on statistical learning theory (SLT) has high generalization ability, especially for dataset with small number of samples in high dimensional space. SVM was originally developed for two-class classification. It is extended to solve multi-class classification problem. In this paper, 57 dimensional feature vectors for MRI image are selected as input for SVM. The segmentation of MRI image based on the multi-classification SVM (MCSVM) is investigated. As our experiment demonstrates, the boundaries of 7 kinds of encephalic tissues are extracted successfully, and it can reach satisfactory generalization accuracy. Thus, SVM exhibits its great potential in image segmentation.

Strict approximate pattern matching with general gaps

In biological sequence analysis, long and frequently occurring patterns tend to be interesting. Data miners designed pattern growth algorithms to obtain frequent patterns with periodical wildcard gaps, where the pattern frequency is defined as the number of pattern occurrences divided by the number of offset sequences. However, the existing definition set does not facilitate further research works. First, some extremely frequent patterns are obviously uninteresting. Second, the Apriori property does not hold; consequently, state-of-the art algorithms are all Apriori-like and rather complex. In this paper, we propose an alternative definition of the number of offset sequences by adding a number of dummy characters at the tail of sequence. With the new definition, these uninteresting patterns are no longer frequent, and the Apriori property holds, hence our Apriori algorithm can mine all frequent patterns with minimal endeavor. Moreover, the computation of the number of offset sequences becomes straightforward. Experiments with a DNA sequence indicate 1) the pattern frequencies under two definition sets have little difference, therefore it is reasonable to replace the existing one with the new one in practice, and 2) our algorithm runs less rounds than the best case of MMP which is based on the existing definition set.