Recursive DBPSO for Computationally Efficient Electronic Nose System

Scene classification from images is a challenging problem in computer vision due to its significant variability of scale, illumination, and view. Recently, Latent Dirichlet Allocation (LDA) model has grown popular in computer vision field, especially in scene labeling and classification. However, the effectiveness of the LDA model for the scene classification has not yet been addressed thoroughly. Especially, there is little experimental evaluation on the model's performance for different types of features. Fusion of multiple types of features is usually necessary in the scene classification due to the complexity of scene images. In this paper, we investigate the effectiveness of the LDA model in scene classification by using 7 types of features (i.e. uniform grid based interest points, Harris corner based interest points, scale invariant feature transform (SIFT), texture, shape, color, and location) and their various combinations. Furthermore, we compare the performance of the LDA model with Support Vector Machine (SVM) classifier. All experiments are performed on the UIUC Sport Scene database. The experiments demonstrate that the performance of the LDA model 1) is significantly lower than the SVM classifier for the scene classification over different types of features; and 2) decreases by fusing multiple features while improvement shown in SVM classifier.

A multi-ensemble method based on deep auto-encoders for fault diagnosis of rolling bearings

In this work, we have reported a novel automatic technique to detect early breast cancer by analyzing breast thermogram. The system comprises mainly of three steps: the breast region segmentation from the original image, feature extraction, and finally classification and performance analysis. In the segmentation phase, we have first removed the background region by applying the Otsu's thresholding method followed by a reconstruction technique. Then the inframammary fold is detected to mark the lower limit of the breast. After that, the upper limit of the breast is identified by discerning the axilla. Finally, the breast region is extracted based on these two limits. In the next phase, we have extracted features from the region of interest to identify the early breast cancer, which is the most crucial and significant step. At the very beginning of the feature extraction process, we have computed the initial feature point image (IFI) for each segmented breast thermogram by applying discrete wavelet transform (DWT). After that, different types of features are extracted from the IFI for the diagnosis of breast cancer. Finally, feed-forward artificial neural network with gradient decent training rule is employed here as a classifier. The major problem in this research is the limited collection of publicly available breast thermal databases. At present, there is only one standard breast thermal database available updated by Visual Lab, Fluminense Federal University, Brazil. We are also developing a new database, which will help the researcher to do research in this field. In this work, we have used 306 breast thermograms of 102 patients collected from Visual Lab. In our proposed system, we found the accuracy of 90.48%, whereas sensitivity and specificity were 87.6% and 89.73% respectively.

Jeffries-Matusita (JM) distance, a transformation of the Bhattacharyya distance, is a widely used measure of the spectral separability distance between the two class density functions and is generally used as a class separability measure. It can be considered to have good potential to be used for evaluation of the effectiveness of a feature in discriminating two classes. The capability of JM distance as a ranking based feature selection technique for binary classification problems has been verified in some research works as well as in our earlier work. It was found by our simulation experiments with benchmark data sets that JM distance works equally well compared to other popular feature ranking methods based on mutual information, information gain or Relief. Extension of JM distance measure for feature ranking in multiclass problems has also been reported in the literature. But all of them are basically rank based approaches which deliver the ranking of the features and do not automatically produce the final optimal feature subset. In this work, a novel heuristic approach for finding out the optimum feature subset from JM distance based ranked feature lists for multiclass problems have been developed without explicitly using any specific search technique. The proposed approach integrates the extension of JM measure for multiclass problems and the selection of the final optimal feature subset in a unified process. The performance of the proposed algorithm has been evaluated by simulation experiments with benchmark data sets in comparison with two other previously developed multiclass JM distance measures (weighted average JM distance and another multiclass extension equivalent to Bhattacharyya bound) and some other popular filter based feature ranking algorithms. It is found that the proposed algorithm performs better in terms of classification accuracy, F-measure, AUC with a reduced set of features and computational cost.

A novel test optimizing algorithm for sequential fault diagnosis

Harmony search (HS), inspired by the music improvisation process, is a new meta-heuristic optimization method and has been used to tackle various optimization problems in discrete and continuous space successfully. However, the standard HS algorithm is not suitable for settling discrete binary problems. To extend HS to solve the binary-coded problems effectively, a novel discrete binary harmony search (DBHS) algorithm is proposed in this paper. A new pitch adjustment rule is developed to enhance the optimization ability of DBHS. Then parameter studies are performed to investigate the properties of DBHS, and the recommended parameter values are given. The results of numerical experiments demonstrate that the proposed DBHS is valid and outperforms the discrete binary particle swarm optimization algorithm and the standard HS.

In emotion recognition, many irrelevant and redundant features will affect recognition results, so feature selection is necessary.Aimed at emotion physiological signal feature selection, this paper proposed with improved discrete binary particle swarm optimization(BPSO) to increase the correct classification rate of emotion state.When recognizing four emotional states with nearest classifier by four physiological signals, the whole correct recognition rate is up to 85%.Experimental results demonstrate that the BPSO is an effective way to emotion physiological signals feature selection.

Item Splitting splits an item into two items rated under two alternative contextual conditions respectively for improving the prediction accuracy of contextual recommendations. To get more specialized rating data, Complex Splitting is proposed to further improve the accuracy of recommendations. The key of the approach is to select multiple contextual conditions for splitting user or item. We translate it into a contextual conditions combinatorial optimization problem based on discrete binary particle swarm optimization (BPSO) algorithm. The item or user is split into two different items or users according to those contextual conditions in optimal combination. We evaluate our algorithm through a real world dataset and the experimental results demonstrate its validity and reliability.

Reducing the volatility in the operation of public transport is still an important consideration when seeking to maintain efficient and stable operations. In this study, stations and intersections are considered as a whole, and a self-coordinated holding strategy is developed based on the fluctuation of passenger flows and the characteristics of signals during off-peak periods. On this basis, a collaborative optimization is carried out in combination with signal priority and speed control. The mathematical model is established with the goal of minimizing passenger travel and enterprise operation costs. A section of the K2 line in Ganzhou City is selected for a case study. The improved discrete binary particle swarm optimization algorithm (DPSO) is then used for the calculations. The simulation and comparison of the arrival time distribution, headway deviation, and travel time index before and after the optimization show that reliability, stability, and efficiency have been improved by the optimization, which verifies the effectiveness of the optimization method.

Interruptible load management plays an important role in maintaining the reliability and security of power systems and reducing price spike of power markets, thus has been widely used in power systems. Interruptible loads dispatch problem with multi-period and multi-interruptible-load is usually a multi-objective, combinatorial optimization problem, and its solution method needs to be carefully investigated. Application of intelligent optimization algorithms to the multi-period interruptible load dispatch problem is discussed in this paper. Multi objectives, such as minimizing the interruption payments and interruption frequency, can be included in the proposed interruptible load dispatch model. The characteristics of different interruptible loads are also considered. Numerical simulations using discrete binary particle swarm optimization (BPSO) and the genetic algorithm (GA) show that the BPSO can obtain better solutions than that of the GA. Thus the BPSO presents better potentials in solving the interruptible load dispatch problem.

In this paper, we propose a formulation of the unit commitment problem based on the profit under the deregulated electricity market(PBUC). We express the unit commitment problem as a mixed integer nonlinear optimization problem in which the objective is to maximize profits for generation company and the decisions are required to meet all kinds of operating constraints. Under the assumption of competitive market and price forecasting, we developed an improved discrete binary particle swarm optimization(PSO) and standard value PSO to solve this PBUC problem iteratively. A generation company with 10 generating units is used to demonstrate the effectiveness of the proposed approach. Simulation results are compared with those obtained from reference method.

As a newly uncovered post-translational modification on the ε-amino group of lysine residue, protein malonylation was found to be involved in metabolic pathways and certain diseases. Apart from experimental approaches, several computational methods based on machine learning algorithms were recently proposed to predict malonylation sites. However, previous methods failed to address imbalanced data sizes between positive and negative samples. In this study, we identified the significant features of malonylation sites in a novel computational method which applied machine learning algorithms and balanced data sizes by applying synthetic minority over-sampling technique. Four types of features, namely, amino acid (AA) composition, position-specific scoring matrix (PSSM), AA factor, and disorder were used to encode residues in protein segments. Then, a two-step feature selection procedure including maximum relevance minimum redundancy and incremental feature selection, together with random forest algorithm, was performed on the constructed hybrid feature vector. An optimal classifier was built from the optimal feature subset, which featured an F1-measure of 0.356. Feature analysis was performed on several selected important features. Results showed that certain types of PSSM and disorder features may be closely associated with malonylation of lysine residues. Our study contributes to the development of computational approaches for predicting malonyllysine and provides insights into molecular mechanism of malonylation.

In this paper, we investigate identification of human subjects from electrocardiogram (ECG) signals. We segment the ECG records into individual heartbeat based on the localization of R wave peaks. Two types of features, namely analytic and appearance features, are extracted to represent the characteristics of heartbeat signal of different subjects. Feature selection is performed to find out significant attributes. We compared the performance of different classification algorithms. To better utilize the advantages of different types of features, we proposed two schemes for data fusion and classification. Our system achieves promising results with 100% correct human identification rate and 98.90% accuracy for heartbeat identification. The proposed framework reveals the potential of employing appearance based analysis in ECG signal, yet demonstrates the advantage of a hierarchical architecture in pattern recognition problems.

ABSTRACTWaste disposal processes and landfill management are crucial subjects in the field of settlement archaeology. Our study is focused on understanding the processes that are connected to the formation of the infills of settlement features and the recycling of the building materials (daub and wood) and waste management. These research questions are addressed through the analysis of plant macroremains, charcoals, phytoliths, starch, micromorphology, phosphates and magnetic susceptibility. The results show the waste character of feature infills, which reflect the economy of the individual households. The composition of the archaeobotanical assemblages is not influenced by the type of feature, and similarities in the plant spectra often emerge in the features located close to each other. The charcoal and dendrochronological analyses suggest that part of the building's oak timber was recycled and later used as fuelwood. The presence of uncharred remains of vegetative and generative parts in wastes was detected by phytolith and starch analyses and the presence of meat and dung of livestock by animal protein analyses. Anomalies in soil phosphate indicate differing amounts of organic matter in individual features. Magnetic susceptibility measurements showed that each feature infill was formed in a one‐time event. The sedimentological analysis demonstrated that the infills of different types of features were similar, but only the infill of Late and Final Bronze Age cultures' specific trench‐like features was compacted. All proxies combined suggest that prior deposition of waste in sunken features first accumulated elsewhere and became mixed/homogenized.

Melanoma is the deadliest form of skin cancer that causes around 75% of deaths worldwide. However, most of the skin cancers can be cured, especially if detected and treated early. Existing approaches have employed various feature extraction methods, where different types of features are used individually for skin image classification which may not provide sufficient information to the classification algorithm necessary to discriminate between classes, leading to sub-optimal performance. This study develops a novel skin image classification method using multi-tree genetic programming (GP). To capture local information from gray and color skin images, Local Binary Pattern is used in this work. In addition, for capturing global information, variation in color within the lesion and the skin regions, and domain-specific lesion border shape features are extracted. GP with a multi-tree representation is employed to use multiple types of features. Genetic operators such as crossover and mutation are designed accordingly in order to select a single type of features at terminals in one tree of the GP individual. The performance of the proposed method is assessed using two skin image datasets having images captured from multiple modalities, and compared with six most commonly used classification algorithms as well as the standard (single-tree) wrapper and embedded GP methods. The results show that the proposed method has significantly outperformed all these classification methods. Being interpretable and fast in terms of the computation time, this method can help dermatologist identify prominent skin image features, specific to a type of skin cancer in real-time situations.