Protein phosphorylation governs cellular signaling, making accurate identification of kinase-specific sites essential for understanding regulatory and disease mechanisms. Although computational approaches have shown promise in inferring kinase specificity, most existing methods primarily rely on local sequence patterns and remain limited in their ability to capture broader contextual information. More critically, experimentally validated kinase-substrate data are inherently scarce and highly imbalanced across kinase groups, substantially restricting the generalization performance of purely discriminative models, especially for underrepresented kinases. To address these challenges, we propose KSDiffusion, a unified framework for kinase-specific phosphorylation site prediction explicitly designed for data-limited and imbalanced regimes. KSDiffusion integrates a protein language model with task-aware conditional diffusion-based generative modeling. Specifically, an ESM-2-based encoder is employed to extract context-aware peptide representations enriched with evolutionary and structural information, while supervised contrastive learning further enhances kinase-specific discriminability in the embedding space. To alleviate data scarcity for rare kinase groups, we introduce a conditional diffusion model, termed KS-DiT, which generates biologically plausible and kinase-consistent synthetic representations that directly support downstream prediction. Comprehensive experiments across kinase groups spanning low-, medium-, and large-data regimes demonstrate that KSDiffusion consistently outperforms representative baseline methods. In particular, substantial improvements are achieved for data-scarce kinase groups, with AUC gains of up to $\sim $15%, while maintaining competitive performance when sufficient training data are available. These results underscore the regime-dependent effectiveness of conditional diffusion-based augmentation and highlight the value of integrating protein language models with task-aware generative modeling for robust kinase-specific phosphorylation site prediction under realistic data constraints.

Constraint-handling techniques and genetic operators are two crucial components in constrained multi-objective evolutionary algorithms (CMOEAs). Recent research in most of CMOEAs has primarily focused on adaptive designs of these components to address various constrained multi-objective optimization problems (CMOPs). However, the evolutionary process of solving a CMOP can involve various characteristics, such as continuity, discreteness, degeneracy, or some combination thereof, necessitating the tailored selection of constraint-handling techniques and genetic operators across different generations. This study conceptualizes these selections as a temporal sequence of constrained handling selection, where the time means the generation number. We argue that discovering the systematic patterns within the sequence based on the historical data of applying different selections significantly improves the performance of CMOEAs in finding Pareto optimal solutions. Based on this conceptualization, we propose a CMOEA with a deep reinforcement learning model for solving CMOPs. Specifically, the deep reinforcement learning model dynamically refines the selection of constraint-handling techniques and genetic operators for upcoming generations by learning from the performance of previous selections, thereby enhancing the predictive accuracy for subsequent selections. Experiments are conducted to validate the performance of the proposed algorithm against nine CMOEAs on thirty-seven benchmark problems and an unmanned aerial vehicle path planning problem. Experimental results show that the proposed algorithm substantially outperforms the compared algorithms regarding the obtained Pareto optimal solutions. Additionally, the results verify that discovering the systematic patterns within the sequence for CMOEAs has a positive impact on solving CMOPs in terms of objective optimization and constraint satisfaction.

Recent studies exploited Large Language Models (LLMs) to autonomously generate heuristics for solving Combinatorial Optimization Problems (COPs), by prompting LLMs to first provide search directions and then derive heuristics accordingly. However, the absence of task-specific knowledge in prompts often leads LLMs to provide unspecific search directions, obstructing the derivation of well-performing heuristics. Moreover, evaluating the derived heuristics remains resource-intensive, especially for those semantically equivalent ones, often requiring omissible resource expenditure. To enable LLMs to provide specific search directions, we propose the Hercules algorithm, which leverages our designed Core Abstraction Prompting (CAP) method to abstract the core components from elite heuristics and incorporate them as prior knowledge in prompts. We theoretically prove the effectiveness of CAP in reducing unspecificity and provide empirical results in this work. To reduce computing resources required for evaluating the derived heuristics, we propose few-shot Performance Prediction Prompting (PPP), a first-of-its-kind method for the Heuristic Generation (HG) task. PPP leverages LLMs to predict the fitness values of newly derived heuristics by analyzing their semantic similarity to previously evaluated ones. We further develop two tailored mechanisms for PPP to enhance predictive accuracy and determine unreliable predictions, respectively. The use of PPP makes Hercules more resource-efficient and we name this variant Hercules-P. Extensive experiments across four HG tasks, five COPs, and eight LLMs demonstrate that Hercules outperforms the state-of-the-art LLM-based HG algorithms, while Hercules-P excels at minimizing required computing resources. In addition, we illustrate the effectiveness of CAP, PPP, and the other proposed mechanisms by conducting relevant ablation studies.

To address the weight coupling problem, certain studies introduced few-shot Neural Architecture Search (NAS) methods, which partition the supernet into multiple sub-supernets. However, these methods often suffer from computational inefficiency and tend to provide suboptimal partitioning schemes. To address this problem more effectively, we analyze the weight coupling problem from a novel perspective, which primarily stems from distinct modules in succeeding layers imposing conflicting gradient directions on the preceding layer modules. Based on this perspective, we propose the Gradient Contribution (GC) method that efficiently computes the cosine similarity of gradient directions among modules by decomposing the Vector-Jacobian Product during supernet backpropagation. Subsequently, the modules with conflicting gradient directions are allocated to distinct sub-supernets while similar ones are grouped together. To assess the advantages of GC and address the limitations of existing Graph Neural Architecture Search methods, which are limited to searching a single type of Graph Neural Networks (Message Passing Neural Networks (MPNNs) or Graph Transformers (GTs)), we propose the Unified Graph Neural Architecture Search (UGAS) framework, which explores optimal combinations of MPNNs and GTs. The experimental results demonstrate that GC achieves state-of-the-art (SOTA) performance in supernet partitioning quality and time efficiency. In addition, the architectures searched by UGAS+GC outperform both the manually designed GNNs and those obtained by existing NAS methods. Finally, ablation studies further demonstrate the effectiveness of all proposed methods.

Low-light image enhancement (LLIE) aims at restoring the color and brightness of underexposed images. Supervised methods suffer from high costs in collecting paired low-normal light images, while unsupervised approaches require intricate loss functions. To tackle these dual challenges, we propose the Trouble-Maker Learning (TML) strategy, which leverages images with normal light as training inputs. TML comprises two core components. Firstly, the Troublemaker Model (TM) generates pseudo low-light images from normal images, thereby alleviating the need for pairwise data and reducing associated costs. Secondly, the Predicting Model (PM) enhances the brightness of pseudo low-light images. Additionally, we integrate an Enhancing Model (EM) to further refine the visual quality of the PM's outputs. In LLIE tasks, it is crucial to capture global element correlations, as this allows for the extraction of more information pertaining to the same object. Convolutional Neural Networks (CNNs) and self-attention mechanisms are not well-suited to this task due to the local CNN operators, and high time complexity, respectively. To address these limitations, we propose Global Dynamic Convolution (GDC) with a time complexity of O(n). Essentially, GDC mimics the partial calculation process of self-attention to establish element-wise correlations. Building upon the GDC module, we develop the UGDC model. Finally, we explore the application of Data Fusion in the field of LLIE. Based on the Retinex theory, we conducted feature-level fusion using low-light images, illumination components and reflection components, which further enhance the performance of the LLIE system. Extensive quantitative and qualitative experiments demonstrate that UGDC, trained with TML and via data fusion, can achieve performance competitive with state-of-the-art approaches on public datasets. The source code of this paper is publicly available at https://github.com/Rainbowman0/TML_LLIE, facilitating reproducibility of the research findings.

Existing online video super-resolution methods utilize implicit memories of previous frames to provide reference information, which have a single memory stream path and are highly dependent on the continuous memory stream. However, video capture in real-world scenes is typically affected by abnormal exposures resulting in sudden changes of lightness thus interrupting the memory stream, while long-term memories suffer from memory vanishing problems during transmission. To address this problem, we propose a novel multi-memory streams based online video super-resolution paradigm that adaptively corrects for abnormal exposures and creates multi-memory streams to accurately converge long-term memories. Specifically, we first propose an exposure detection-correction module, which utilizes optical flow overfitting property and temporal lightness information to detect and correct abnormal exposures to avoid interruption of memory streams. In addition, we propose a dynamic-static decoupled alignment strategy, which can adaptively select the alignment method based on pixel displacement, thus accurately aggregating past long-term memories to create multiple memory streams. Further, we propose an adaptive memory fusion module to mine complementary information between multiple memory streams to solve the memory vanishing problem. Extensive experimental results show that our method outperforms existing video super-resolution methods on complex exposure datasets. We also conduct detailed ablation experiments to analyze and validate our contributions.

Unsupervised visible-infrared person re-identifi-cation (Unsupervised VI-ReID) aims to learn discriminative identity features under the large modality gap without any labeled data. Currently, the state-of-the-art methods optimize cross-modality differences by using contrastive learning as the underlying paradigm. However, they neglect the problem of modality divergence during the cross-modality optimization process. This problem means that the interclass instances between the cross-modality intraclass gaps can make cross-modality intraclass instances difficult to get closer to each other in the feature space due to the effect of contrastive learning on these interclass instances. To alleviate the negative impact of the modality divergence problem, we propose an adversarial aware learning (ADAL) framework to explore the instances that generate modal divergence and adversarially optimize these explored instances. Specifically, on the one hand, we explore the optimization directions of each cluster during the cross-modality optimization process, and the cluster centroids generating positive optimization are facilitated, while the others generating negative optimization are penalized. On the other hand, we further consider the instance-level optimization process, which increases the affinities of the positive instance pairs with large cross-modality gaps to further improve the centroid-level optimization. Extensive experiments conducted on the visible-infrared person Re-ID datasets show that the proposed method is used as a universally applicable plug-in module to add the existing unsupervised VI-ReID methods, which outperforms the existing state-of-the-art approaches.

Despite deep neural networks have made outstanding achievements in many static tasks, when faced with a continuous stream of data, they suffer from catastrophic forgetting since the previous data is usually inaccessible. Stored data or generative model is commonly used for maintaining the model performance but with memory utilization and privacy safety issues. Prototype-based methods address these issues by keeping only one prototype for each class but with limitations in its ability to trade-off the model stability and plasticity. In this paper, a novel exemplar-free class-incremental learning method is proposed which improves the stability of the representation learning and the decision boundary to a great degree. First, based on the results of our exploration into the impact of the batch normalization (BN) layer on representation learning, we propose to remove the BN layer (RBNL) in the incremental training phase to improve the stability of model representation learning. Then, to further maintain the feature space, we design the prototype mixing (PM), which expands the deep features by randomly and linearly combining prototypes of the old classes to generate hybrid prototypes with composite labels for fine-tuning the fully connected layer. Experimental results on three benchmark datasets, CIFAR-100, TinyImageNet, and ImageNet, show that our proposed method can effectively balance the stability and plasticity of the model, and outperforms the state-of-the-art works.

Cartesian Genetic Programming (CGP) has often been applied for Neural Architecture Search (NAS). However, the performance of CGP is less than ideal when searching for architectures with limited computing resources. To better facilitate NAS with limited computing resources, this paper proposes a crossover operator, a light-weighted age mechanism, and two adaptive mutation operators as the novel components in our Efficient Cartesian Genetic Programming (ECGP) method. To assess the performance of ECGP, we conduct extensive experiments on three text classification task datasets. The experimental results demonstrate that ECGP outperforms other NAS methods, requiring only hundreds of fitness evaluations to find architectures with competitive accuracy compared with human-designed models. Additionally, the ECGP-evolved architectures are shown as converging fast and stably, and having high-level transferability with merely a 1-2% accuracy drop. Ablation studies demonstrate the effectiveness of the proposed operators and age mechanism, and identify GRU as the most critical function in the text classification task. Finally, we summarize three design principles observed from the ECGP-evolved architectures that are in line with human-design strategies. To the best of our knowledge, this work introduces the first attention-derived NAS benchmark for the text classification task.

The surprisingly popular decision in social science fields is a wisdom of the crowd technique that taps into the expert minority opinion within a crowd, which has been demonstrated to be remarkably effective for multiple questions. Most of the existing PSO variants construct the exemplars by solely using fitness, which could be viewed as the democratic approaches or methods. However, the democratic methods tend to highlight the most popular opinion, not necessarily the most correct, which might lead the population into a local trapping region in the scenarios of swarm intelligent computing and evolutionary computation. This paper proposes a method to implement the surprisingly popular decision in PSO to facilitate the exemplar construction, cooperating with the dynamic topology maintenance. The proposed PSO variant is called the Surprisingly Popular Algorithm-based Comprehensive Adaptive Topology Learning Particle Swarm Optimization (SPA-CatlePSO). By using the dynamic topological connection and surprisingly popular decision strategy, the proposed SPA-CatlePSO could adjust the degree of small world topology, mimicking the mechanism of knowledge conversion in the crowd, and guide the direction of the exploitation by constructing exemplars with the largest surprisingly popular degree. We evaluate the proposed SPA-CatlePSO on the full CEC2014 benchmark suite and compare its validity with OLPSO, TSLPSO, ASDPSO, HCLPSO, OptBees and L-shade. The experimental results show that the SPA-CatlePSO algorithm is competitive with the most advanced swarm-based intelligent algorithms.

Particle swarm optimizations (PSOs) are drawing extensive attention from both research and engineering fields due to their simplicity and powerful global search ability. However, there are two issues needing to be improved: one is that the classical PSO converges slowly; the other is that classical PSO tends to result in premature convergence, especially for multi-modal problems. This paper attempts to address these two issues. Firstly, to improve the convergent efficiency, this paper proposes an asymptotic predicting model of the globally-optimal solution, which is used to predict the global optimum based on extracting the features reflecting the evolutionary trend. The predicted global optimum is then taken as the third exemplar, in a way similar to the individual historical best solution and the swarm historical best solution in guiding the evolutionary process of other particles. To reduce the probability that the population is trapped into a local optimum due to the premature phenomenon, this paper proposes an adaptive mutation strategy, which is used to help the trapped particles to escape away from the local optimum by using the extended non-uniform mutation operator. Finally, we combine the two entities to develop a globally-optimal prediction-based adaptive mutation particle swarm optimization (GPAM-PSO). In numerical experimental parts, we compare the proposed GPAM-PSO with 11 existing PSO variants by using 22 benchmark problems of 30-dimensions and 100-dimensions, respectively. Numerical experiments demonstrate that the proposed GPAM-PSO could improve the accuracy and efficiency remarkably, which means that the combination of the globally-optimal prediction-based search and the adaptive mutation strategy could accelerate the convergence and reduce premature phenomenon effectively. Generally speaking, GPAM-PSO performs most efficiently and robustly. Moreover, the performance on an engineering problem demonstrates the practical application of the proposed GPAM-PSO algorithm.

MOTIVATION: Mislabeled samples often appear in gene expression profile because of the similarity of different sub-type of disease and the subjective misdiagnosis. The mislabeled samples deteriorate supervised learning procedures. The LOOE-sensitivity algorithm is an approach for mislabeled sample detection for microarray based on data perturbation. However, the failure of measuring the perturbing effect makes the LOOE-sensitivity algorithm a poor performance. The purpose of this article is to design a novel detection method for mislabeled samples of microarray, which could take advantage of the measuring effect of data perturbations. RESULTS: To measure the effect of data perturbation, we define an index named perturbing influence value (PIV), based on the support vector machine (SVM) regression model. The Column Algorithm (CAPIV), Row Algorithm (RAPIV) and progressive Row Algorithm (PRAPIV) based on the PIV value are proposed to detect the mislabeled samples. Experimental results obtained by using six artificial datasets and five microarray datasets demonstrate that all proposed methods in this article are superior to LOOE-sensitivity. Moreover, compared with the simple SVM and CL-stability, the PRAPIV algorithm shows an increase in precision and high recall. AVAILABILITY: The program and source code (in JAVA) are publicly available at http://ccst.jlu.edu.cn/CSBG/PIVS/index.htm

Ant colony optimization (ACO) has widely been applied to solve combinatorial optimization problems in recent years. There are few studies, however, on its convergence time, which reflects how many iteration times ACO algorithms spend in converging to the optimal solution. Based on the absorbing Markov chain model, we analyze the ACO convergence time in this paper. First, we present a general result for the estimation of convergence time to reveal the relationship between convergence time and pheromone rate. This general result is then extended to a two-step analysis of the convergence time, which includes the following: 1) the iteration time that the pheromone rate spends on reaching the objective value and 2) the convergence time that is calculated with the objective pheromone rate in expectation. Furthermore, four brief ACO algorithms are investigated by using the proposed theoretical results as case studies. Finally, the conclusions of the case studies that the pheromone rate and its deviation determine the expected convergence time are numerically verified with the experiment results of four one-ant ACO algorithms and four ten-ant ACO algorithms.