A Fuzzy Deep Model Based on Fuzzy Restricted Boltzmann Machines for High-dimensional Data Classification

In this paper, we propose aself-constructing neural fuzzy network with dynamic-form symbiotic evolution (SCNFN-DSE) for solving various problerns. A novel hybrid learning approach, which consists of the self-clustering algorithm (SCA) and the dynamic-form symbiotic evolution (DSE), is proposed for adjusting the parameters of neural fuzzy networks. First, the proposed SCA is used to identify a pazsimonious internal structure. The SCA is an online clustering method and is a distance-based connectionist clustering method. Second, the proposed DSE uses the sequential-search based dynamic evolution (SSDE) method. The better chromosomes will be initially generated while the better mutation points will be determined for performing dynamic-mutation. Simulation results have shown that 1) the SCNFN-DSE model converges quickly; 2) the SCNFNDSE model requires a small number of population sizes; 3) the SCNFN-DSE model construct only 4 fuzzy models in every generation.

Image classification serves as the cornerstone of computer vision, traditionally achieved through discriminative models based on deep neural networks. Recent advancements have introduced classification methods derived from generative models, which offer the advantage of zero-shot classification. However, these methods suffer from two main drawbacks: high computational overhead and inferior performance compared to discriminative models. Inspired by the coordinated cognitive processes of rapid-slow pathway interactions in the human brain during visual signal recognition, we propose the Diffusion-Based Discriminative Model Enhancement Framework (DBMEF). This framework seamlessly integrates discriminative and generative models in a training-free manner, leveraging discriminative models for initial predictions and endowing deep neural networks with rethinking capabilities via diffusion models. Consequently, DBMEF can effectively enhance the classification accuracy and generalization capability of discriminative models in a plug-and-play manner. We have conducted extensive experiments across 17 prevalent deep model architectures with different training methods, including both CNN-based models such as ResNet and Transformer-based models like ViT, to demonstrate the effectiveness of the proposed DBMEF.Specifically, the framework yields a 1.51% performance improvement for ResNet-50 on the ImageNet dataset and 3.02% on the ImageNet-A dataset. In conclusion, our research introduces a novel paradigm for image classification, demonstrating stable improvements across different datasets and neural networks.

Relief demand forecasting is vital to the success of disaster relief operations, but it is associated with challenges including insufficient training samples, incomplete and imprecise inputs, and inaccurate demands. This article presents a co-evolutionary fuzzy deep transfer learning (CoFDTL) method for relief demand forecasting where different types of disasters (e.g., earthquake, typhoon, and flood) are considered as different tasks. CoFDTL consists of three stages. First, a deep fuzzy learning model is used to learn latent representation of the shared inputs of all tasks. Second, a co-evolutionary algorithm is used to simultaneously learn task-specific features and the shared regressor. Third, the shared regressor is re-trained based on the best solutions obtained for different tasks in the second stage. Experiments demonstrate that CoFDTL exhibits significant performance improvements over the selected popular fuzzy learning, deep learning, and transfer learning models. This article also reports the application of CoFDTL to two real-world disasters in China, 2018. The proposed CoFDTL that integrates fuzzy deep learning, transfer learning, and co-evolutionary learning can be used for many other complex multi-task transfer learning problems with insufficient samples and uncertain information.

Complex distributed parameter systems (DPSs) are prevalent in numerous industrial processes. However, the nonlinear spatiotemporal dynamics inherent in DPS present significant challenges for accurate modeling. In this paper, an innovative online three-dimensional (3D) fuzzy multi-output support vector regression learning method is proposed for DPS modeling. The proposed method employs spatial fuzzy basis functions from the 3D fuzzy model as kernel functions, enabling direct construction of a comprehensive fuzzy rule base. Parameters C and ε in the 3D fuzzy model adaptively adjust according to data sequence variations, effectively responding to system dynamics. Furthermore, a stochastic gradient descent algorithm has been implemented for real-time updating of learning parameters and bias terms. The proposed method was validated through two typical DPS and an actual rotary hearth furnace industrial system. The experimental results show the effectiveness of the proposed modeling method.

Transfer learning has been widely adopted for few-shot classification. Recent studies reveal that obtaining good generalization representation of images on novel classes is the key to improving the few-shot classification accuracy. To address this need, we prove theoretically that leveraging ensemble learning on the base classes can correspondingly reduce the true error in the novel classes. Following this principle, a novel method named Ensemble Learning with Multi-Order Statistics (ELMOS) is proposed in this paper. In this method, after the backbone network, we use multiple branches to create the individual learners in the ensemble learning, with the goal to reduce the storage cost. We then introduce different order statistics pooling in each branch to increase the diversity of the individual learners. The learners are optimized with supervised losses during the pre-training phase. After pre-training, features from different branches are concatenated for classifier evaluation. Extensive experiments demonstrate that each branch can complement the others and our method can produce a state-of-the-art performance on multiple few-shot classification benchmark datasets.

Medical supervised masked autoencoder: Crafting a better masking strategy and efficient fine-tuning schedule for medical image classification

Objective: This paper proposes a method for classifying sleep–wakefulness and estimating sleep parameters using nasal pressure signals applicable to a continuous positive airway pressure (CPAP) device. Approach: In order to classify the sleep–wakefulness states of patients with sleep-disordered breathing (SDB), apnea-hypopnea and snoring events are first detected. Epochs detected as SDB are classified as sleep, and time-domain- and frequency-domain-based features are extracted from the epochs that are detected as normal breathing. Subsequently, sleep–wakefulness is classified using a support vector machine (SVM) classifier in the normal breathing epoch. Finally, four sleep parameters—sleep onset, wake after sleep onset, total sleep time and sleep efficiency—are estimated based on the classified sleep–wakefulness. In order to develop and test the algorithm, 110 patients diagnosed with SDB participated in this study. Ninety of the subjects underwent full-night polysomnography (PSG) and twenty underwent split-night PSG. The subjects were divided into 50 patients of a training set (full/split: 42/8), 30 of a validation set (full/split: 24/6) and 30 of a test set (full/split: 24/6). Main results: In the experiments conducted, sleep–wakefulness classification accuracy was found to be 83.2% in the test set, compared with the PSG scoring results of clinical experts. Furthermore, all four sleep parameters showed higher correlations than the results obtained via PSG (r ⩾ 0.84, p < 0.05). In order to determine whether the proposed method is applicable to CPAP, sleep–wakefulness classification performances were evaluated for each CPAP in the split-night PSG data. The results indicate that the accuracy and sensitivity of sleep–wakefulness classification by CPAP variation shows no statistically significant difference (p < 0.05). Significance: The contributions made in this study are applicable to the automatic classification of sleep–wakefulness states in CPAP devices and evaluation of the quality of sleep.

Early identification of lung cancer using questionnaire-based data offers a low-cost, non-invasive pathway to assist clinical decision-making. However, such datasets often contain redundant, noisy, and imbalanced attributes that limit the performance of traditional classifiers. This study introduces a hybrid LSTM-GRU framework optimized using a Grey Wolf-Whale Optimization (GWO-WOA) algorithm for hyperparameter tuning and Binary Particle Swarm Optimization (BPSO) for feature selection. Two public lung cancer datasets sourced from the Kaggle repository were employed: the first comprising 309 samples and the second containing 3000 samples. For both datasets, the preprocessing pipeline included missing-value imputation, categorical encoding, outlier removal, and z-score normalization to ensure feature consistency. Datasets were then split into 70%, 20%, and 10% subsets for training, validation, and testing, respectively. BPSO effectively selected the most informative features that contribute to accurate diagnosis. At the same time, GWO-WOA refined key hyperparameters, such as the learning rate, hidden units, and layer depth, of the hybrid architecture. Experimental results demonstrate the superior performance of the proposed GWO-WOA-LSTM-GRU model, achieving 100.00% accuracy, precision, recall, and F1-score on the 309-sample dataset, and 99.33% accuracy/F1 (precision: 99.34%, recall: 99.33%) on the 3000-sample dataset. In comparison, tuned single models-LSTM, GRU, CNN, and SVM-achieved accuracies ranging from 77.42 to 98.33%. These findings confirm that integrating metaheuristic optimization and hybrid recurrent networks enhances the robustness and generalization capabilities of lung cancer classification systems across diverse datasets, offering a reliable tool for early detection and clinical risk stratification.

This paper analyzes the role of dictionary selection in Sparse Representation-based Classification (SRC). While SRC introduces interesting results in the field of classification, its performance is highly limited by the number of training samples to form the classification matrix. Different studies addressed this issue by using a more compact representation of the training data in order to achieve higher classification speed and accuracy. Representative selection methods which are analyzed in this paper include Metaface dictionary learning, Fisher Discriminative Dictionary Learning (FDDL), Sparse Modeling Representative Selection (SMRS), and random selection of the training samples. The first two methods build their own dictionaries via an optimization process while the other two methods select the representatives directly from the original training samples. These methods, along with the original method which uses all training samples to form the classification matrix, were examined on two face datasets and one digit dataset. The role of feature extraction was also studied using two dimensionality reduction methods, down-sampling and random projection. The results show that the FDDL method leads to the best classification accuracy followed by the SMRS method as the second best. On the other hand, the SMRS method requires a much smaller learning time which makes it more appropriate for dynamic situations where the dictionary is regularly updated with new samples. The accuracy of the Metaface dictionary learning method was specifically less than the other two methods. As expected, using all the training samples as the dictionary resulted in the best recognition rates in all the datasets but the classification times for this approach were far larger than the required time using any of the three dictionary learning methods.

This commentary to the recent article by Musso et al. (2013) discusses issues related to model fitting, comparison of classification accuracy of generative and discriminative models, and two (or more) cultures of data modeling. We start by questioning the extremely high classification accuracy with an empirical data from a complex domain. There is a risk that we model perfect nonsense perfectly. Our second concern is related to the relevance of comparing multilayer perceptron neural networks and linear discriminant analysis classification accuracy indices. We find this problematic, as it is like comparing apples and oranges. It would have been easier to interpret the model and the variable (group) importance’s if the authors would have compared MLP to some discriminative classifier, such as group lasso logistic regression. Finally, we conclude our commentary with a discussion about the predictive properties of the adopted data modeling approach.

A nondestructive method for evaluation of tomato quality was proposed to recognize the different ripening stages of tomatoes and to predict their hardness. Computer vision and electronic nose were used, and data fusion technology was used to establish the decision model of multi‐attribute information fusion. Fisher discriminant analysis (FDA) model and support vector classifier (SVC) model for tomato grading were established. Partial least squares (PLS) model and support vector regression (SVR) model were used for predicting the hardness. Comparing with the single detection technique, the combined system which fused computer vision and electronic nose, achieved better results. In the combined system: with respect to the SVC model, classification accuracy was 96.38% for the training set, 94.20% for the prediction set; while to the FDA model, classification accuracy was 93.48% for the training set, 85.51% for the prediction set; to the PLS model, the correlation coefficient of prediction (Rp) was 94.09% and the root mean square error of prediction (RMSEP) was 2.33 N; to the SVR model, the Rp was 95.14% and RMSEP was 0.03 N. It was observed that more robust and better prediction performance on tomato detection were achieved by using fusion information.Practical applicationsComputer vision, electronic nose, data fusion, and pattern recognition were used for the evaluation of tomato ripeness and hardness. Computer vision captured the color information of tomatoes and electronic nose responded to the smell of tomatoes. After data fusion, information became more abundant and the discrimination models also got better performance. It proved the possibility of nondestructive method for the evaluation of tomatoes quality during storage. So it is feasible to apply this method to detect tomato quality in factory.

Two cooperative ant colonies for feature selection using fuzzy models

Based on Variable Universe fuzzy control technology with good adaptive and robust capability, a new excitation control strategy for DFIG(doubly fed induction generator) without accurate generator mathematical model is proposed in this paper. Full fuzzy excitation control model is established and analyzed in detail. the whole wind generation system model Was established, simulation results indicate that the proposed methods is efectiveness for maintaining optimum power operating, even during generation system characters varying.

The accurate and non-destructive assessment of fruit ripeness is essential for post-harvest sorting and quality management. This study evaluated a meta-inspired classification framework integrating partial least squares discriminant analysis (PLS-DA) with support vector machines (SVMs) trained on latent variables (sSVM) or on class probabilities (pSVM) derived from multiple PLS-DA components. Two kiwiberry varieties, ‘Geneva’ and ‘Weiki’, were analyzed using variety-specific and combined datasets. Performance was assessed in calibration and prediction using accuracy, F05, Cohen’s kappa, precision, sensitivity, specificity, and likelihood ratios. Conventional PLS-DA provided reasonably good classification, but pSVM models, particularly those with an RBF kernel (pSVM_R), consistently outperformed other approaches and ensured higher stability across all datasets. Unlike sSVMs, which were prone to overfitting, pSVM_R models achieved the highest accuracy of 92.4–96.9%, Cohen’s kappa of 84.8–93.9%, and precision of 89.1–94.2%, clearly surpassing both score-based SVM and PLS-DA. Contrasting tendencies were observed between cultivars: ‘Geneva’ models improved during prediction, while ‘Weiki’ models declined, especially in specificity. Combined datasets provided greater stability but slightly reduced peak performance than single-variety models. These findings highlight the value of probability-enriched stacking models for non-invasive ripeness discrimination, suggesting that adaptive or hybrid strategies may further enhance generalization across diverse cultivars.

This article explores the use of single trial EEG signals to predict the voluntary movements of single hand and two hands. During single-hand movements, three kinds of task, grasping, releasing, and holding were considered. The tasks considered during two-hand movements are left and right grasping, left and right releasing, and holding. The subject performs the tasks spontaneously without waiting for and responding to any external cues. In addition, a neural adaptive noise canceller is developed that accomplishes eye blinks suppression. The neural adaptive filter is here implemented by means of a three-layer feed-forward neural network. The feature vectors are formed from the three channels (Fz, C3, and F3). We employ the multilayer perceptron (MLP) with back-propagation learning algorithm and Radial Basis Function (RBF) network with stochastic gradient learning rule for discriminating different patterns of the EEG signals. In the classical approach to RBF and MLP network implementation, the number of hidden units is predetermined. It, usually, results in too many hidden units. To overcome this drawback, we develop an enhanced resource-allocating network (RAN) for discriminating the EEG patterns. These networks start with no hidden units and grow by allocating new hidden units based on the novelty in the EEG signals, which arrive sequentially. The results of this analysis show that the neural networks would be able to detect the movements of a single hand and two hands with an average classification accuracy of 98.82% and 96.40%, respectively. Moreover, the RAN provides a reduction in the training epochs as compared to the MLP and RBF networks. This work represents a promising approach to control prosthesis device.