Fuzzy rough feature selection (FRFS) effectively alleviates the curse of dimensionality by eliminating redundant and irrelevant features, thereby improving model generalization. However, most existing algorithms focus on minimizing classification uncertainty, even though lower uncertainty does not necessarily imply stronger class discrimination or improved classification performance. This challenges the common assumption that uncertainty alone sufficiently captures feature relevance in pattern classification tasks. To bridge this gap, we propose a Margin-Aware Fuzzy Rough Feature Selection (MAFRFS) framework that explicitly incorporates structural characteristics of class distributions, namely, within-class compactness and between-class separability, into the feature evaluation process. By integrating margin-based structural cues with fuzzy rough uncertainty modeling, MAFRFS effectively guides the selection toward more separable and discriminative feature subsets. Extensive experiments reported on 23 publicly available datasets demonstrate that MAFRFS is highly scalable and more effective than FRFS. Algorithms developed under MAFRFS consistently outperform some state-of-the-art feature selection algorithms.

Gait Emotion Recognition (GER) is an emerging task within Human Emotion Recognition. Skeleton-based GER requires discriminative spatial and temporal features. However, current methods primarily focus on capturing spatial topology information but fail to effectively learn temporal features from long-distance frames. Moreover, these methods are mostly sensitive to the order of sampled sequences, resulting in significant accuracy drops when sequences are randomly sampled. In order to obtain a more robust and comprehensive spatial-temporal representation of gait, we introduce the Graph-Transformer architecture into GER for the first time, proposing a novel framework named GaitCycFormer. Specifically, we designed a Cycle Position Encoding (CPE) based on the gait cycle, which explicitly segments any gait sequence into more manageable periodic units, to enhance temporal feature modeling. Additionally, we incorporate a bi-level Transformer, consisting of an Intra-cycle Transformer and an Inter-cycle Transformer to capture local and global temporal information within each gait cycle and between gait cycles respectively. Experiments demonstrate that our GaitCycFormer achieves state-of-the-art performance on popular datasets, and proves to be more reliable and robust.

Micro-expressions are brief and subtle facial movements. Unlike macro-expressions, they are difficult to control and can reflect true emotion. Therefore, they are valuable in criminal investigation, medical care and other applications. Micro-expression recognition refers to the emotion classification of micro-expression samples. Previous approaches frequently relied on image sequences as inputs and ignored the fact that micro-expressions are activated only in local areas, introducing irrelevant noise. Additionally, some methods solely employed traditional graph models without fully exploring the complex spatiotemporal relationships between different facial regions and frames. To address this issue, we propose the method with a Graph Transformer for micro-expression recognition to more effectively learn the interrelations between facial regions and frames, thereby obtaining more discriminative features. Specifically, we develop a novel Graph Transformer with a learnable adjacency matrix for spatiotemporal learning, which better learns long-range dependencies and adaptively integrates implicit information in the graph. We select appropriate facial landmarks and calculate the optical-flow-based feature to serve as input. Finally, experiments conducted on relevant datasets have demonstrated the effectiveness of our method.

Multimodal aspect-oriented sentiment classification (MABSC) task has garnered significant attention, which aims to identify the sentiment polarities of aspects by combining both language and vision information. However, the limited multimodal data in this task has become a big gap for the vision-language multimodal fusion. While large-scale vision-language pretrained models have been adapted to multiple tasks, their use for MABSC task is still in a nascent stage. In this work, we present an attempt to use the instruction tuning paradigm to MABSC task and leverage the ability of large vision-language models to alleviate the limitation in the fusion of textual and image modalities. To tackle the problem of potential irrelevance between aspects and images, we propose a plug-and-play selector to autonomously choose the most appropriate instruction from the instruction pool, thereby reducing the impact of irrelevant image noise on the final sentiment classification results. We conduct extensive experiments in various scenarios and our model achieves state-of-the-art performance on benchmark datasets, as well as in few-shot settings.

Black-box attacks pose a significant challenge due to the restricted access to target model information, hindering the generation of impactful adversarial samples. This paper introduces a method that combines sparse attacks and meta-learning to alleviate the issue of low success rates in black-box attacks. SAM leverages the knowledge transfer capabilities inherent in meta-learning to augment the transferability of adversarial samples. The method integrates meta-learning with gradient-based attack techniques, effectively transforming the approach into a white-box attack. By aggregating multiple sampled models, SAM enhances the stability of adversarial samples. During meta-testing, simulated black-box attacks help mitigate gradient discrepancies across diverse models, consequently enhancing transferability. To further improve sparsity and preserve transferability, SAM incorporates a projection strategy that selectively sparsifies global adversarial perturbations. Experimental evaluations conducted on two image datasets substantiate SAM’s superiority in terms of both sparsity and attack success rate. Ablation experiments confirm the effectiveness of integrating meta-learning into the proposed method. SAM extends the applicability of generated adversarial samples, advancing the domain of adversarial attacks in scenarios with limited target model information. The proposed approach exhibits promise in enhancing the success rate of attacks while preserving sparsity, contributing to the broader understanding of black-box attacks and their implications.

Image captioning aims to generate fluent and accurate descriptions for images. To evaluate the quality of captions, various metrics have been proposed. However, current metrics only assess captions at sequence-level, which overwhelms the distinctions between each token. Thus, existing objectives tend to treat each token equally, assigning them with identical weights in loss functions. Intuitively, key words in a caption carry the primary information and contribute more than other words to sequence-level metrics. They should be distinguished and weighted more during training. In this work, we propose to explicitly measure each word and guide the model to focus more on key words in captions. Firstly, we devise token-level CIDEr (CIDEr-T) as a new metric to quantify the importance of each word, by decomposing the sequence-level CIDEr into token-level granularity. CIDEr-T maintains consistency with CIDEr and shows the distinctions between tokens. Thus, we engage CIDEr-T scores of each token as their unique weights in the raw loss functions, which can bridge the gap between training and evaluation.

Unbalanced classification is an essential machine learning task, which has attracted widespread attention from both the academic and industrial communities due mainly to its broad applications. Evolutionary computation (EC) has contributed greatly to unbalanced classification. However, to the best of our knowledge, there have not been any comprehensive investigations on the strengths and weaknesses of alternative EC methods in addressing various challenging problems in unbalanced classification. This article reviews the literature which utilize EC techniques for unbalanced classification, with the aim of revealing the contributions of EC to unbalanced classification, providing an overview of recent advances, and identifying limitations of existing works. In addition, we present a series of real-world applications, and identify open challenges as well as possible research directions for the future.

Facial expression feature extraction suffers from high inter-subject variations caused by identity-related personal attributes. The extracted expression features are consistently entangled with other identity-related features, which has an influence on related facial expression tasks such as recognition and editing. To achieve high-quality expression features, a Disentangled Variational Autoencoder (DisVAE) is proposed to disentangle expression and identity features. The identity features are removed from the facial features via facial image reconstruction firstly, and then the remaining features represent expression components. Extensive experiments on three public datasets have shown that the proposed DisVAE can effectively disentangle expression and identity features, and extract expression features without the interfere of identity attributes. The high-quality expression features improve the performance of facial expression recognition and can be well applied to facial expression editing.

Question-controlled Text-aware Image Captioning (Qc-TextCap), is the task of generating a distinctive scene text aware caption according to several personalized questions when given an image. However, due to the diversity of visual scene, it is hard for current Optical Character Recognition (OCR) systems to extract complete scene text sentences from the images. Besides, existing works are limited in their use of question features and visual features. In this paper, we propose a Multimodal Transformer plus Scene text clustering and Cross modal attention (MTSC) to tackle the above challenges. We devise scene text clustering to group relevant scene text pieces which are detected as separate results by current OCR systems. To better utilize the information in questions and images, we design cross modal attention to enrich the features of both modalities. We extensively evaluate our model on the two Qc-TextCap datasets and superior results are achieved when comparing to state-of-the-art approaches.

Chinese Spelling Check (CSC) aims to detect and correct Chinese spelling errors. Most Chinese spelling errors are the misuse of semantically, phonetically or graphically similar characters. Previous state-of-the-art works on the CSC task pursue transitions from misspelled sentences to correct sentences directly. However, it is difficult to force the current CSC methods to find the correct answer at one run. Thus, we propose a simple and effective method for CSC task by making fully use of the trained model to generate multiple candidate sentences and simply ranking to select the best, in which no additional training and parameters are required. The experimental results show that our approach outperforms previous methods and achieves the state-of-the-art performances.

Unbalanced classification has attracted widespread interest because of its broad applications. However, due to mainly the uneven class distribution, constructed classifiers are usually biased toward the majority class, and thereby perform terribly on the minority class. Unfortunately, the minority class is often the class of interest in many real-world applications. High dimensionality often further degrades the classification performance, making it more complicated to address the class imbalance issue. Genetic programming (GP) has been applied to construct classifiers, which can simultaneously select good-quality features to improve the classification performance. To handle the class imbalance issue, cost-sensitive GP classifiers treat the minority class as being more important than the majority class, but this may cause an accuracy decrease in overlapping areas where the prior probabilities of the two classes are almost the same. To date, most cost-sensitive classification methods have not been specifically investigated how the impacts of overlapping areas on cost-sensitive classifiers can be avoided. In this study, we propose a new cost-sensitive GP method, where rough set theory is employed to detect overlapping areas before training cost-sensitive classifiers for classification with unbalanced high-dimensional data. The proposed method is compared with 46 popular classification methods, including 10 GP methods and 36 non-GP methods on 14 datasets that are unbalanced and high dimensional. The experimental results indicate that the proposed method performs better than the compared methods in almost all cases.

Retrieval-based methods have been proven effective in many NLP tasks.Previous methods use representations from the pre-trained model for similarity search directly.However, the sentence representations from the pre-trained model like BERT perform poorly in retrieving semantically similar sentences, resulting in poor performance of the retrieval-based methods.In this paper, we propose KNN-EC, a simple and efficient non-parametric emotion classification (EC) method using nearest neighbor retrieval.We use BERT-whitening to get better sentence semantics, ensuring that nearest neighbor retrieval works.Meanwhile, BERTwhitening can also reduce memory storage of datastore and accelerate retrieval speed, solving the efficiency problem of the previous methods.KNN-EC average improves the pre-trained model by 1.17 F1-macro on two emotion classification datasets.

TextCaps, also known as image captioning with reading comprehension, is the task of automatically describing images with both visual objects and scene text in them. It is more challenging than conventional image captioning since it requires models to read scene text and cover them in generated captions. Recently, various models have achieved excellent results on this task. However, existing approaches are limited in their use of spatial relationships between all the visual entities (both visual objects and scene text). Captions from these models can hardly describe the explicit spatial relations between the text and relevant objects. In this paper, we propose a Spatial Relationship Incorporated Multimodal Transformer (SRIMT) to generate spatial-aware captions. Firstly, we construct a spatial graph to fully explore the spatial relationships between all the visual entities. Then, we present a novel spatially aware self-attention layer which is the core of our model to incorporate spatial relationship. Through this layer, attention of two visual entities is considered only when they are connected in the spatial graph. Furthermore, each head in our multi-head attention module is designed to focus on only one type of spatial relation defined by the specific graph. Compared with fully-connected transformer-based architectures, our model can learn the spatial relationships of a visual scene more explicitly instead of dispersing attention among all visual entities. Strong spatial relations between OCR tokens and corresponding objects are established with our model. We extensively evaluate our model on the TextCaps dataset and superior results are achieved when comparing to state-of-the-art approaches. More remarkably, we improve CIDEr score from 93.0 to 95.8.

In sentiment analysis, it is concerned recently that a significant portion of subjective sentences across different domains do not contain explicit sentiment words but still convey clear subjective sentiment, which is known as implicit sentiment. However, currently available methods do not perform well on implicit sentiment analysis. In most cases implicit sentiment can be inferred from the context. To address this issue, we propose the ContextBert model, which employs implicit-sentiment-query attention to find the contextual information related to the target sentence to enhance implicit sentiment analysis. Experimental results show that our method achieves state-of-the-art performance on SMP2019-ECISA and EmoContext-Implicit (reconstructed based on EmoContext, to enrich the implicit sentiment analysis dataset).

High-dimensional unbalanced classification is challenging because of the joint effects of high dimensionality and class imbalance. Genetic programming (GP) has the potential benefits for use in high-dimensional classification due to its built-in capability to select informative features. However, once data are not evenly distributed, GP tends to develop biased classifiers which achieve a high accuracy on the majority class but a low accuracy on the minority class. Unfortunately, the minority class is often at least as important as the majority class. It is of importance to investigate how GP can be effectively utilized for high-dimensional unbalanced classification. In this article, to address the performance bias issue of GP, a new two-criterion fitness function is developed, which considers two criteria, that is, the approximation of area under the curve (AUC) and the classification clarity (i.e., how well a program can separate two classes). The obtained values on the two criteria are combined in pairs, instead of summing them together. Furthermore, this article designs a three-criterion tournament selection to effectively identify and select good programs to be used by genetic operators for generating offspring during the evolutionary learning process. The experimental results show that the proposed method achieves better classification performance than other compared methods.

In classification, when class overlap is intertwined with the issue of class imbalance, it is often challenging to discover useful patterns because of an ambiguous boundary between the majority class and the minority class. This becomes more difficult if the data is high-dimensional. To date, very few pieces of work have investigated how the class overlap issue can be effectively addressed or alleviated in classification with high-dimensional unbalanced data. In this paper, we propose a new genetic programming based method, which is able to automatically and directly detect borderline instances, in order to address the class overlap issue in classification with high-dimensional unbalanced data. In the proposed method, each individual has two trees to be trained together based on different classification rules. The proposed method is examined and compared with baseline methods on high-dimensional unbalanced datasets. Experimental results show that the proposed method achieves better classification performance than the baseline methods in almost all cases.

Person re-identification (re-ID), aiming to recognize a person of interest across different cameras, is a well-known challenge because of the vast variation in human poses. Existing pose-driven re-ID methods utilize keypoints, which are considered fine-grained local information, inadequate to capture global pose characteristics. In this paper we propose multi-granularity pose fusion network (MGPFNet), to address the pose variation problem. In particular we propose the use of view information in re-ID. A pose in MGPFNet is multi-granular containing the coarse-grained view and fine-grained keypoint information. Two encoding methods, namely pose concatenation encoding and pose independent encoding, are proposed to incorporate both coarse-grained views information and fine-grained keypoints information while alleviating the impact of pose variations. Extensive experiments on three benchmark datasets with views demonstrates the effectiveness of MGPFNet which outperforms state-of-the-art methods in most cases, confirming the benefit of using views in re-ID.

Event detection based on sentiment time series, which describe the trend of users' emotions or attitudes towards specific topics over time, has been widely applied in the analysis of social network or text mining. Most of the contributions directly generate time series sequences by classifiers. However, due to the missing corpus labels or the limited performance of the classifier, such generated sentiment time series may not correspond to the actual values, especially when the sentiment value changes drastically, called extreme value. We propose a new method to calibrate sentiment times series for event detection based on evaluation on a sampling dataset. Theoretical analysis of the calibration method is explicated, and it is proved that the sampling error of the performance indicators can be limited to a minimal range for extreme values, thus sentiment value error can be reduced. Experiments on simulated datasets and real-world datasets illustrate the effectiveness and robustness of our method.

Class imbalance is an unavoidable issue in many real-world applications. Learning from unbalanced data, classifiers are often biased toward the majority class, while the minority class is important as well (even more important in many cases). How the issue of class imbalance is addressed becomes more challenging if a classification task further encounters the high dimensionality issue. This paper proposes a new genetic programming (GP) approach to high-dimensional unbalanced classification. A new classification mechanism is proposed for GP to improve its classification performance. This new classification mechanism is independent of a classification threshold to separate the majority class and the minority class. The effectiveness of the proposed method is examined on seven high-dimensional unbalanced datasets. Experimental results indicate that the proposed GP method often performs better than other GP methods that use a fitness function to solve the issue of class imbalance, in terms of classification performance and training time.

High-dimensionality and class imbalance represent two main challenges in classification. Recently, there is a growing number of datasets exhibiting the characteristics of the combination of the class imbalance and high-dimensionality. Genetic programming (GP) has been successfully applied to solve high-dimensional classification tasks. However, most existing GP methods may also suffer from a performance bias if the class distribution is unbalanced. Using fitness functions for cost adjustment is one of the most important methods in GP to address the class imbalance issue. This paper develops new fitness functions in GP to address the class imbalance issue in classification with high-dimensional unbalanced data. Two fitness functions are proposed to increase the performance of the traditional accuracy measures, and one fitness function is proposed to approximate Area Under Curve (AUC) with the goal to save the training time. Experiments on six high-dimensional unbalanced datasets show the better performance of the proposed fitness functions, compared to existing fitness functions.

Pruning can be used in Convolutional Neural Networks(CNNs) to reduce the consumption of computations. In the iteration of pruning, a fixed ratio of filters or weights are pruned after evaluating their importance according to a certain criterion, which has the risk of leading to mis-pruning when the distribution of the importance is imbalanced. We propose a strategy to assist pruning for CNNs at filter level. Firstly, in order to evaluate importance and non-importance about filters in the network, we design two fuzzy membership functions respectively, then we evaluate all filters by their membership function value, finally, α-cut set is employed to determine the ones to be pruned . Experimental results demonstrate the effectiveness of our strategy, since it can obtain more compression than the popular pruning algorithms at similar accuracy level.

Label distribution learning (LDL) is a newly arisen learning paradigm to deal with label ambiguity problems, which can explore the relative importance of different labels in the description of a particular instance. Although some existing LDL algorithms have achieved better effectiveness in real applications, most of them typically emphasize on improving the learning ability by manipulating the label space, while ignoring the fact that irrelevant and redundant features exist in most practical classification learning tasks, which increase not only storage requirements but also computational overheads. Furthermore, noises in data acquisition will bring negative effects on the generalization performance of LDL algorithms. In this paper, we propose a novel algorithm, i.e., Latent Semantics Encoding for Label Distribution Learning (LSE-LDL), which learns the label distribution and implements feature selection simultaneously under the guidance of latent semantics. Specifically, to alleviate noise disturbances, we seek and encode discriminative original physical/chemical features into advanced latent semantic features, and then construct a mapping from the encoded semantic space to the label space via empirical risk minimization. Empirical studies on 15 real-world data sets validate the effectiveness of the proposed algorithm.

Multi-task learning is an effective approach to extract task-invariant features by leveraging potential information among related tasks, which improves the performance of a single task. Most existing work simply divides the whole model into shared and private spaces. Unfortunately, there is no explicit mechanism to prevent the two spaces from merging information from each other. As a result, the shared space may be mixed with task-specific features, while the private space may extract some task-invariant features. To alleviate the problem mentioned, in this paper, we propose a bidirectional language models based multi-task learning method for text classification. More specifically, we add language modelling as an auxiliary task to the private part, aiming to enhance its ability to extract task-specific features. In addition, to promote the shared part to learn common features, a loss constraint via uniform label distribution is introduced to the shared part. Finally, put task-specific features and taskinvariant features together in a weighted addition way to form the final representation, and it is then fed to the corresponding softmax layer. We do experiments on the FDU-MTL dataset which consists of 16 different text classification tasks. The experimental results show that our approach outperforms other typical methods.

The bilateral filter method is a nonlinear filter with spatial averaging without smoothing edges. It has shown to be an effective image denoising technique. Denoising performance using the bilateral filter is affected by the filter parameters, which are image dependent and require experimental trials. We propose an automatic and effective PSO-based method of parameters selection for the bilateral filter in image denoising. Intensity domain parameter δr and the radius parameter d are optimized by the PSO algorithm, in which SSIM (structural similarity index) is employed in fitness function. We firstly compare our approach with other four classical filtering methods at different types and levels of noise. We also compare the denoising performance with different values of the parameter δd. Experimental results on three sets of color images have shown that the proposed method of parameter selection outperformed the other filtering methods in denoising standard test images corrupted by different types and levels of noise.

Image segmentation is an important issue in computer vision. The methods based on Fuzzy C-Means (FCM) algorithms have gained success. However, these approaches deal with each pixel as a separate object, which will ignore the spatial information among these pixels. This paper proposes an approach which combines the Fuzzy C-Means algorithm and Graph Cut Theory both for gray and color image segmentation. We adopt the Turbopixel algorithm to split the color image into varied small regions called superpixels for presegmentation and extract color histogram features from the superpixels. Based on color histogram feature, we use FCM to make the original clusters. Then we build a graph model, and use maximum flow algorithm to get the minimum cut, namely the initial segmentation result of the image. Finally, we use a recursive process to achieve the result of image segmentation. The key point of our approach is building a great graphical model and utilizing the existing binary segmentation model to solve the multi-value segmentation. Experimental results show that our approach can obtain good segmentation results comparing with FCM only under different parameters setup and binary segmentation.

Generalisation is one of the most important performance measures for any learning algorithm, no exception to Genetic Programming (GP). A number of works have been devoted to improve the generalisation ability of GP for symbolic regression. Methods based on a reliable estimation of generalisation error of models during evolutionary process are a sensible choice to enhance the generalisation of GP. Structural risk minimisation (SRM), which is based on the VC dimension in the learning theory, provides a powerful framework for estimating the difference between the generalisation error and the empirical error. Despite its solid theoretical foundation and reliability, SRM has seldom been applied to GP. The most important reason is the difficulty in measuring the VC dimension of GP models/programs. This paper introduces SRM, which is based on an empirical method to measure the VC dimension of models, into GP to improve its generalisation performance for symbolic regression. The results of a set of experiments confirm that GP with SRM has a dramatical generalisation gain while evolving more compact/less complex models than standard GP. Further analysis also shows that in most cases, GP with SRM has better generalisation performance than GP with bias-variance decomposition, which is one of the state-of-the-art methods to control overfitting.

Genetic programming(GP) has become an increasingly hot issue in evolutionary computation due to its extensive application. Anomaly detection in crowded scenes is also a hot research topic in computer vision. However, there are few contributions on using genetic programming to detect abnormalities in crowded scenes. In this paper, we focus on anomaly detection in crowded scenes with genetic programming. We propose a new method called Multi-Frame LBP Difference(MFLD) based on Local Binary Patterns(LBP) to extract pixel-level features from videos without additional complex preprocessing operations such as optical flow and background subtraction. Genetic programming is employed to generate an anomaly detector with the extracted data. When a new video is coming, the detector can classify every frame and localize the abnormality to a single-pixel level in realtime. We validate our approach on a public dataset and compare our method with other traditional algorithms for video anomaly detection. Experimental results indicate that our method with genetic programming performs better in detecting abnormalities in crowded scenes.

This paper proposes an approach which combines the Decision Theoretic Rough Set model (DTRS) and Fuzzy C-Means(FCM) algorithm to perform color image segmentation. The FCM algorithm has the limitation that it requires the initialization of cluster centroids and the number of clusters. In this paper, the DTRS model is applied to color image segmentation for the purpose of clustering validity analysis which could overcome the defect of the FCM algorithm. Firstly, we adopt the Turbopixel algorithm to split the color image into many small regions called superpixels for presegmentation. Based on color image color histogram feature extraction we use Bhattacharyya coefficient to measure the similarity between superpixels, which is in preparation for clustering validity analysis. It is our focus that we will obtain cluster centroids and the number of clusters using FCM. Our approach is according to the hierarchical clustering validity analysis algorithm using DTRS model. Finally, the FCM algorithm is utilized to achieve the result of color image segmentation. Experimental results show that the DTRS-based preprocessing approach can obtain better segmentation results than other improved FCM approaches such as ant colony algorithm or histogram thresholding approach.

Feature selection is a multi-objective problem, where the two main objectives are to maximize the classification accuracy and minimize the number of features. However, most of the existing algorithms belong to single objective, wrapper approaches. In this work, we investigate the use of binary particle swarm optimization (BPSO) and probabilistic rough set (PRS) for multi-objective feature selection. We use PRS to propose a new measure for the number of features based on which a new filter based single objective algorithm (PSOPRSE) is developed. Then a new filter-based multi-objective algorithm (MORSE) is proposed, which aims to maximize a measure for the classification performance and minimize the new measure for the number of features. MORSE is examined and compared with PSOPRSE, two existing PSO-based single objective algorithms, two traditional methods, and the only existing BPSO and PRS-based multi-objective algorithm (MORSN). Experiments have been conducted on six commonly used discrete datasets with a relative small number of features and six continuous datasets with a large number of features. The classification performance of the selected feature subsets are evaluated by three classification algorithms (decision trees, Naïve Bayes, and k-nearest neighbors). The results show that the proposed algorithms can automatically select a smaller number of features and achieve similar or better classification performance than using all features. PSOPRSE achieves better performance than the other two PSO-based single objective algorithms and the two traditional methods. MORSN and MORSE outperform all these five single objective algorithms in terms of both the classification performance and the number of features. MORSE achieves better classification performance than MORSN. These filter algorithms are general to the three different classification algorithms.

Dimension reduction plays an important role in many classification tasks. In this work, we propose a new filter dimension reduction algorithm (PSOPRSE) using binary particle swarm optimisation and probabilistic rough set theory. PSOPRSE aims to maximise a classification performance measure and minimise a newly developed measure reflecting the number of attributes. Both measures are formed by probabilistic rough set theory. PSOPRSE is compared with two existing PSO based algorithms and two traditional filter dimension reduction algorithms on six discrete datasets of varying difficulty. Five continues datasets including a large number of attributes are discretised and used to further examine the performance of PSOPRSE. Three learning algorithms, namely decision trees, nearest neighbour algorithms and naive Bayes, are used in the experiments to examine the generality of PSOPRSE. The results show that PSOPRSE can significantly decrease the number of attributes and maintain or improve the classification performance over using all attributes. In most cases, PSOPRSE outperforms the first PSO based algorithm and achieves better or much better classification performance than the second PSO based algorithm and the two traditional methods, although the number of attributes is slightly large in some cases. The results also show that PSOPRSE is general to the three different classification algorithms.

Although many particle swarm optimization (PSO) based feature subset selection methods have been proposed, most of them seem to ignore the difference of feature subset selection problems and other optimization problems. We analyze the search process of a PSO based wrapper feature subset selection algorithm and find that characteristics of feature subset selection can be used to optimize this process. We compare wrapper and filter ways of evaluating features and define the domain knowledge of feature subset selection problems and we propose a fast wrapper feature subset selection algorithm based on PSO employed the domain knowledge of feature subset selection problems. Experimental results show that our method can work well, and the new algorithm can improve both the running time and the classification accuracy.

Fuzzy rough sets are generalization of rough sets to handle fuzziness and uncertainty existed in data. The decision rough set model (DTRS) is the kind of probabilistic rough set model with proper cost functions. We combine fuzzy sets with decision-theoretic rough set (DTRS) theory and propose a new equation by employing the fuzzy membership functions to change the posterior probability calculating method containing in the expected losses expression of the DTRS model. Thus we can derive the new decision rules. With different user demands, varied thresholds in DTRS are firstly set to decide with probability an object can be classified to the positive region. For different fuzzy membership functions, we propose the method to select the appropriate one based on user-demand in our new fuzzy rough set model. Experiments on different datasets show that different membership functions do result in different classification performances when threshold has been set in advance. With our method it can give guidelines on appropriate fuzzy membership functions selection for improving classification accuracy.

Based on binary particle swarm optimisation (BPSO) and information theory, this paper proposes two new filter feature selection methods for classification problems. The first algorithm is based on BPSO and the mutual information of each pair of features, which determines the relevance and redundancy of the selected feature subset. The second algorithm is based on BPSO and the entropy of each group of features, which evaluates the relevance and redundancy of the selected feature subset. Different weights for the relevance and redundancy in the fitness functions of the two proposed algorithms are used to further improve their performance in terms of the number of features and the classification accuracy. In the experiments, a decision tree (DT) is employed to evaluate the classification accuracy of the selected feature subset on the test sets of four datasets. The results show that with proper weights, two proposed algorithms can significantly reduce the number of features and achieve similar or even higher classification accuracy in almost all cases. The first algorithm usually selects a smaller feature subset while the second algorithm can achieve higher classification accuracy.

Feature selection has the two main objectives of minimising the classification error rate and the number of features. Based on binary particle swarm optimisation (BPSO), we develop two novel multi-objective feature selection frameworks for classification, which are multi-objective binary PSO using the idea of non-dominated sorting (NSBPSO) and multi-objective binary PSO using the ideas of crowding, mutation and dominance (CMDBPSO). Four multi-objective feature selection methods are then developed by applying mutual information and entropy as two different filter evaluation criteria in each of the proposed frameworks. The proposed algorithms are examined and compared with a single objective method on eight benchmark data sets. Experimental results show that the proposed multi-objective algorithms can evolve a set of solutions that use a smaller number of features and achieve better classification performance than using all features. In most cases, NSBPSO achieves better results than the single objective algorithm and CMDBPSO outperforms all other methods mentioned above. This work represents the first study on multi-objective BPSO for filter-based feature selection.

In this paper, we propose a novel approach to clustering noisy and complex data sets based on the eXtend Classifier Systems (XCS). The proposed approach, termed XCSc, has three main processes: (a) a learning process to evolve the rule population, (b) a rule compacting process to remove redundant rules after the learning process, and (c) a rule merging process to deal with the overlapping rules that commonly occur between the clusters. In the first process, we have modified the clustering mechanisms of the current available XCS and developed a new accelerate learning method to improve the quality of the evolved rule population. In the second process, an effective rule compacting algorithm is utilized. The rule merging process is based on our newly proposed agglomerative hierarchical rule merging algorithm, which comprises the following steps: (i) all the generated rules are modeled by a graph, with each rule representing a node; (ii) the vertices in the graph are merged to form a number of sub-graphs (i.e. rule clusters) under some pre-defined criteria, which generates the final rule set to represent the clusters; (iii) each data is re-checked and assigned to a cluster that it belongs to, guided by the final rule set. In our experiments, we compared the proposed XCSc with CHAMELEON, a benchmark algorithm well known for its excellent performance, on a number of challenging data sets. The results show that the proposed approach outperforms CHAMELEON in the successful rate, and also demonstrates good stability.

Aiming at the incipient fault detection of power transformer, the paper proposes a novel fuzzy classification by evolutionary neural network. The method models the membership fuctions of all fuzzy sets by utilizing a three-layer feedforward neural network, and trains a group of neural networks by combining the modified Evolutionary Strategy with Levenberg-Marquardt optimization method in order to accelerate convergence and avoid falling into local minima. Thus each trained neural network denotes an "expert" model. The classification results obtained from all "expert" models are integrated according to the absolute-majority-voting rule. A lot of samples are tested, and the testing results demonstrate that the novel method is much better in neural network structure, classification accuracy, generalization capability, fault-tolerance ability and robustness that then other traditional methods.