ABSTRACT Constrained multi‐objective optimization problems widely exist in real‐world applications, yet remain challenging due to the coexistence of multiple conflicting objectives and constraints. This study proposes a two‐archive constrained multi‐objective optimization algorithm with two‐stage weak cooperation, named BWC‐TAA. First, based on the two‐archive framework, a weak cooperation interaction mechanism is designed, in which the CA and DA evolve independently. They only merge offspring during the update stage to select high‐quality individuals, thereby preventing solutions from being confined to the parent population range. Second, a two‐stage evolutionary strategy is introduced to dynamically adjust the optimization objectives, enabling CA and DA to adopt different strategies in different phases. Finally, BWC‐TAA is evaluated on benchmark constrained problems against eight state‐of‐the‐art algorithms. The experimental results demonstrate that BWC‐TAA significantly outperforms the compared algorithms in terms of convergence and diversity indicators.

Client selection has been widely considered in Federated Learning (FL) to reduce communication overhead while ensuring proper convergence performance. Due to data heterogeneity in FL, a representative subset of participants should take into account both intra-and inter-client diversity. While existing works usually emphasize on one of them, this paper proposes a VDSV (client selection based on Value Density and Secondary Verification) framework, which optimizes the client selection strategy from both sides. Therein, intra-and inter-client diversity are respectively measured based on a designed client data score as well as gradient distance and direction. Afterwards, a client selection model is established based on a proposed metric, called client value density. Besides, a secondary validation method is developed to dynamically tweak the current client selection and model aggregation strategies. The general idea of the above design is based on the theoretical convergence analysis and the observation that the client contribution to the global model can get changed throughout the learning process. The experimental results demonstrate that VDSV can achieve higher convergence rates and ensure comparable model performance. In specific, our method can reduce the communication rounds by an average of 37.88%, which saves noticeable communication overhead.

ABSTRACT Large‐scale multi‐objective optimization problems are widely used in expert systems and applications, which mainly consider the simultaneous optimization of multiple conflicting objectives under large‐scale decision variables. Existing methods typically classify decision variables as either convergence‐ or diversity‐related, neglecting their inherent characteristics and thus failing to balance convergence and diversity effectively. In order to address above issues, this article proposes a Large‐scale Multi‐objective Dual‐population Co‐evolutionary Algorithm based on decision variable boundary penalty (LMDCA). The proposed algorithm first uses a boundary penalty based cross decision variable analysis method to quantitatively analyze the decision variables, which can quantify the contribution values of convergence variables on different objective functions for grouping. Then, according to different variable groups, different optimization strategies are adopted to more accurately approximate the Pareto front of each objective. Subsequently, the convergence and diversity populations were constructed by combining the dual‐population co‐evolutionary framework, in which three strategies of directional restriction of mating choice, environmental selection and information compensation were designed for co‐interaction within each of the populations to ensure the integrity of population evolution information. We conducted extensive comparisons with current algorithms on multiple benchmark datasets and real‐world problems. The experimental results show that the proposed algorithm is superior to the compared algorithms and exhibits strong competitiveness in practical applications.

Numerous studies have shown that in-depth mining of correlations between multi-modal features can help improve the accuracy of cross-modal data analysis tasks. However, the current image description methods based on the encoder-decoder framework only carry out the interaction and fusion of multi-modal features in the encoding stage or the decoding stage, which cannot effectively alleviate the semantic gap. In this paper, we propose a Deep Fusion Transformer (DFT) for image captioning to provide a deep multi-feature and multi-modal information fusion strategy throughout the encoding to decoding process. We propose a novel global cross encoder to align different types of visual features, which can effectively compensate for the differences between features and incorporate each other’s strengths. In the decoder, a novel cross on cross attention is proposed to realize hierarchical cross-modal data analysis, extending complex cross-modal reasoning capabilities through the multi-level interaction of visual and semantic features. Extensive experiments conducted on the MSCOCO dataset prove that our proposed DFT can achieve excellent performance and outperform state-of-the-art methods. The code is available athttps://github.com/weimingboya/DFT.