A Projection Pursuit framework for supervised dimension reduction of high dimensional small sample datasets

Applying virtual sample generation and ensemble modeling for improving the spectral diagnosis of cancer

Cancer diagnosis plays a key role in facilitating treatment and improving survival rates of patients. The combination of near-infrared (NIR) spectroscopy with data-driven algorithms offers a rapid and cost-effective approach for such a task. Due to the limitations of objective cases, the number of tumor samples is usually smaller, and the resulting dataset exhibit the issues of class imbalance, which has a more serious impact on the performance of diagnostic models. To deal with class imbalance and improve the sensitivity, this work investigates the feasibility of NIR spectroscopy combined with virtual sample generation (VSG) as well as ensemble strategy for developing diagnostic models. Based on preliminary experiment, several learning algorithms such as discriminant analysis (DA) and partial least square-discriminant analysis (PLS-DA) are screened out as algorithms for constructing prediction models. Three algorithms of VSG including synthetic minority oversampling technique (SMOTE), Borderline-SMOTE and adaptive synthetic sampling (ADASYN) are used for experiment. A fixed sample subset composed of 27 cancer samples and 54 normal samples are hold out as the test set. Three training sets containing 5, 10, 25 minority class samples and 54 majority class samples are used for model development. The experimental result indicates that overall, with PLS-DA algorithm, all VSG approaches can significantly improve the sensitivity of cancer diagnosis for all cases of training sets with different minority samples, but ADASYN performs the best. It reveals that the integration of NIR, PLS-DA, and ADASYN is a promising tool package for developing diagnosis methods.

Novel manifold learning based virtual sample generation for optimizing soft sensor with small data

Due to the extremely complex mechanism and strong non-linear characteristics of industrial processes, data-driven soft sensor technologies play a key role in the intelligent measurement of process industries. However, the information of the collected process data in the steady stage is quite limited and unreliable, causing the small sample problem. As a result, it becomes an intractable challenge to catch the nature of the process and build accurate soft sensor models. To solve this problem, this paper proposes a novel manifold learning based virtual sample generation method (Isomap-VSG) to generate feasible virtual samples in the information gaps for supplementing the original small sample space. To find data sparse regions reasonably, one kind of manifold learning methods called Isomap is used to visualize process data with high dimension. Then virtual samples can be generated by the interpolation method and extreme learning machine. The simulation results on a standard dataset and a real-world application demonstrate that, compared with other advanced methods, the proposed Isomap-VSG method can achieve better performance in terms of generating feasible virtual samples and improving the accuracy of soft sensor models using limited samples.

Machine learning in classification and identification of nonconventional vegetables

Vegetables are important in economic, social, and nutritional matters in both the Brazilian and international scenes. Hence, some researches have been carried out in order to encourage the production and consumption of different species such as nonconventional vegetables. These vegetables have an added value because of their nutritional quality and nostalgic appeal due to the reintroduction of these species. For this reason, this article proposes the use of the machine learning technique in the construction of models for supervised classification and identification in an experiment with five leafy special of nonconventional vegetables (Tropaeolum majus, Rumex acetosa, Stachys byzantina, Lactuca cf. indica e Pereskia aculeata) assessing the characteristics of the macro and micro nutrients. In order to evaluate the classifiers' performance, the cross-validation procedure via Monte Carlo simulation was considered to confirm the model. In ten replications, the success and error rates were obtained, considering the false positive and false negative rates, sensibility, and accuracy of the classification method. Thus, it was concluded that the use of machine learning is viable because it allows the classification and identification of nonconventional vegetables using few nutritional attributes and obtaining a success rate of over 89% in most of the classifiers tested.

Optimized Mahalanobis-Taguchi System for High-Dimensional Small Sample Data Classification

The Mahalanobis–Taguchi system (MTS) is a multivariate data diagnosis and prediction technology, which is widely used to optimize large sample data or unbalanced data, but it is rarely used for high-dimensional small sample data. In this paper, the optimized MTS for the classification of high-dimensional small sample data is discussed from two aspects, namely, the inverse matrix instability of the covariance matrix and the instability of feature selection. Firstly, based on regularization and smoothing techniques, this paper proposes a modified Mahalanobis metric to calculate the Mahalanobis distance, which is aimed at reducing the influence of the inverse matrix instability under small sample conditions. Secondly, the minimum redundancy-maximum relevance (mRMR) algorithm is introduced into the MTS for the instability problem of feature selection. By using the mRMR algorithm and signal-to-noise ratio (SNR), a two-stage feature selection method is proposed: the mRMR algorithm is first used to remove noise and redundant variables; the orthogonal table and SNR are then used to screen the combination of variables that make great contribution to classification. Then, the feasibility and simplicity of the optimized MTS are shown in five datasets from the UCI database. The Mahalanobis distance based on regularization and smoothing techniques (RS-MD) is more robust than the traditional Mahalanobis distance. The two-stage feature selection method improves the effectiveness of feature selection for MTS. Finally, the optimized MTS is applied to email classification of the Spambase dataset. The results show that the optimized MTS outperforms the classical MTS and the other 3 machine learning algorithms.

A novel gene selection algorithm for cancer classification using microarray datasets

Background

Microarray datasets are an important medical diagnostic tool as they represent the states of a cell at the molecular level. Available microarray datasets for classifying cancer types generally have a fairly small sample size compared to the large number of genes involved. This fact is known as a curse of dimensionality, which is a challenging problem. Gene selection is a promising approach that addresses this problem and plays an important role in the development of efficient cancer classification due to the fact that only a small number of genes are related to the classification problem. Gene selection addresses many problems in microarray datasets such as reducing the number of irrelevant and noisy genes, and selecting the most related genes to improve the classification results.

Methods

An innovative Gene Selection Programming (GSP) method is proposed to select relevant genes for effective and efficient cancer classification. GSP is based on Gene Expression Programming (GEP) method with a new defined population initialization algorithm, a new fitness function definition, and improved mutation and recombination operators. . Support Vector Machine (SVM) with a linear kernel serves as a classifier of the GSP.

Results

Experimental results on ten microarray cancer datasets demonstrate that Gene Selection Programming (GSP) is effective and efficient in eliminating irrelevant and redundant genes/features from microarray datasets. The comprehensive evaluations and comparisons with other methods show that GSP gives a better compromise in terms of all three evaluation criteria, i.e., classification accuracy, number of selected genes, and computational cost. The gene set selected by GSP has shown its superior performances in cancer classification compared to those selected by the up-to-date representative gene selection methods.

Conclusion

Gene subset selected by GSP can achieve a higher classification accuracy with less processing time.

Impacts of urban land use on the spatial distribution of groundwater pollution, Harbin City, Northeast China

Groundwater is the major available water resources in Harbin City, China, but rapid urbanization has adversely affected it in recent years and groundwater quality is deteriorating. In this study, the groundwater pollution indexes of groundwater monitoring wells were analyzed by using the projection pursuit dynamic clustering model and its spatial distribution was interpolated by using the modified local polynomial interpolation method. The "searchlight" shape model was used to extract the types of land use within the scope of each monitoring well in order to evaluate the relationship between the spatial distribution of groundwater pollution and land use. The correlation between the groundwater pollution indexes and land use was evaluated using Kendall's tau-b analysis. The groundwater pollution index is generally high. It has obvious spatial distribution characteristics and it mainly occurs in DaoW-NanG district and the eastern part of DongL-XiangF district. The results of Kendall's tau-b analysis show that the groundwater pollution indexes have a good correlation with land use. Residential and commercial areas, industrial area and agricultural area greatly influence the distribution of groundwater pollution. Moreover, the main impact indicators of groundwater pollution in the study area are NH₄⁺, NO₃^- and CaCO₃.

Novel Online Dimensionality Reduction Method with Improved Topology Representing and Radial Basis Function Networks

This paper presents improvements to the conventional Topology Representing Network to build more appropriate topology relationships. Based on this improved Topology Representing Network, we propose a novel method for online dimensionality reduction that integrates the improved Topology Representing Network and Radial Basis Function Network. This method can find meaningful low-dimensional feature structures embedded in high-dimensional original data space, process nonlinear embedded manifolds, and map the new data online. Furthermore, this method can deal with large datasets for the benefit of improved Topology Representing Network. Experiments illustrate the effectiveness of the proposed method.