Complex & Intelligent Systems最新文献

This paper addresses the economic lot and delivery scheduling problem (ELDSP) within three-echelon supply chains, focusing on the complexities of demand uncertainty, limited shelf-life of products, and sequence-dependency of setups. We develop a novel mixed-integer non-linear programming (MINLP) model for a supply chain comprising one supplier, multiple manufacturers with flexible flow shop (FFS) production systems, and multiple retailers, all operating over a finite planning horizon. The common cycle (CC) strategy is adopted as the synchronization policy. Our model employs fuzzy set theory, particularly the “Me measure,” to effectively handle the retailers’ demand uncertainty. Our findings indicate that total supply chain costs escalate with an increase in demand, final components’ holding costs, and sequence-dependent setup costs, but decrease with increasing production rates. Furthermore, while total costs are significantly sensitive to changes in demand, they are relatively insensitive to fluctuations in sequence-dependent setup times. The models developed offer valuable managerial insights for optimizing costs in synchronized multi-stage supply chains, aiding managers in making informed decisions about production lot sizes and delivery schedules under both deterministic and fuzzy demand scenarios. Additionally, the proposed models bridge key research gaps and provide robust decision-making tools for cost optimization, enhancing supply chain synchronization in practical settings.

Navigating mobile robots in crowded environments poses a significant challenge and is essential for the coexistence of robots and humans in future intelligent societies. As a pragmatic data-driven approach, deep reinforcement learning (DRL) holds promise for addressing this challenge. However, current DRL-based navigation methods have possible improvements in understanding agent interactions, feedback mechanism design, and decision foresight in dynamic environments. This paper introduces the model inductive bias enhanced deep reinforcement learning (MIBE-DRL) method, drawing inspiration from a fusion of data-driven and model-driven techniques. MIBE-DRL extensively incorporates model inductive bias into the deep reinforcement learning framework, enhancing the efficiency and safety of robot navigation. The proposed approach entails a multi-interaction network featuring three modules designed to comprehensively understand potential agent interactions in dynamic environments. The pedestrian interaction module can model interactions among humans, while the temporal and spatial interaction modules consider agent interactions in both temporal and spatial dimensions. Additionally, the paper constructs a reward system that fully accounts for the robot’s direction and position factors. This system's directional and positional reward functions are built based on artificial potential fields (APF) and navigation rules, respectively, which can provide reasoned evaluations for the robot's motion direction and position during training, enabling it to receive comprehensive feedback. Furthermore, the incorporation of Monte-Carlo tree search (MCTS) facilitates the development of a foresighted action strategy, enabling robots to execute actions with long-term planning considerations. Experimental results demonstrate that integrating model inductive bias significantly enhances the navigation performance of MIBE-DRL. Compared to state-of-the-art methods, MIBE-DRL achieves the highest success rate in crowded environments and demonstrates advantages in navigation time and maintaining a safe social distance from humans.

The transformation and upgrading of traditional supply chain models through digital technology receive widespread attention from the fields of circular economy, manufacturing, and sustainable development. Enterprises need to choose a digital supply chain partner (DSCP) during the process of digital transformation in uncertain and sustainable environments. Thus, the research constructs an innovative decision methodology for selecting the optimal DSCP to achieve digital transformation. The proposed methodology is propounded based upon the entropy measure, generalized Dombi operators, integrated weight-determination model, and complex proportional assessment (COPRAS) method under spherical fuzzy circumstances. Specifically, a novel entropy measure is proposed for measuring the fuzziness of spherical fuzzy (SF) sets, while generalized Dombi operators are presented for fusing SF information. The related worthwhile properties of these operators are discussed. Further, an integrated criteria weight-determination model is presented by incorporating objective weights obtained from the SF entropy-based method and subjective weights from the SF best worst method. Afterward, an improvement of the COPRAS method is proposed based on the presented generalized Dombi operators with SF information. Lastly, the practicability and validity of the proposed methodology are verified by an empirical study that selects an appropriate DSCP for a new energy vehicle enterprise to finish the goal of digital transformation. The sensitivity and comparative analysis are carried out to illustrate the stability, reliability, and superiority of the propounded methodology from multiple perspectives. The results and conclusions indicate that the propounded method affords a synthetic and systematic uncertain decision-making framework for identifying the optimal DSCP with incomplete weight information.

Large language models have achieved outstanding performance on various downstream tasks with their advanced understanding of natural language and zero-shot capability. However, they struggle with knowledge constraints, particularly in tasks requiring complex reasoning or extended logical sequences. These limitations can affect their performance in question answering by leading to inaccuracies and hallucinations. This paper proposes a novel framework called KnowledgeNavigator that leverages large language models on knowledge graphs to achieve accurate and interpretable multi-hop reasoning. Especially with an analysis-retrieval-reasoning process, KnowledgeNavigator searches the optimal path iteratively to retrieve external knowledge and guide the reasoning to reliable answers. KnowledgeNavigator treats knowledge graphs and large language models as flexible components that can be switched between different tasks without additional costs. Experiments on three benchmarks demonstrate that KnowledgeNavigator significantly improves the performance of large language models in question answering and outperforms all large language models-based baselines.

Sketch-based cross-domain visual data retrieval is the process of searching for images or 3D models using sketches as input. Achieving feature alignment is a significantly challenging task due to the high heterogeneity of cross-domain data. However, the alignment process faces significant challenges, such as domain gap, semantic gap, and knowledge gap. The existing methods adopt different ideas for sketch-based image and 3D shape retrieval tasks, one is domain alignment, and the other is semantic alignment. Technically, both tasks verify the accuracy of extracted features. Hence, we propose a method based on the global feature correlation and the feature similarity for multiple sketch-based cross-domain retrieval tasks. Specifically, the data from various modalities are fed into separate feature extractors to generate original features. Then, these features are projected to the shared subspace. Finally, domain consistency learning, semantic consistency learning, feature correlation learning and feature similarity learning are performed jointly to make the projected features modality-invariance. We evaluate our method on multiple benchmark datasets. Where the MAP in Sketchy, TU-Berlin, SHREC 2013 and SHREC 2014 are 0.466, 0.473, 0.860 and 0.816. The extensive experimental results demonstrate the superiority and generalization of the proposed method, compared to the state-of-the-art approaches. The in-depth analyses of various design choices are also provided to gain insight into the effectiveness of the proposed method. The outcomes of this research contribute to advancing the field of sketch-based cross-domain visual data retrieval and are expected to be applied to a variety of applications that require efficient retrieval of cross-domain domain data.

As a carrier of multi-industrial technology integration and the key to industrial competition, the intelligent connected vehicle (ICV) has been taken seriously around the world. However, as a fast-growing emerging industry, its development process varies greatly from place to place. Hence, the merits and demerits are analyzed for the development of the ICV industry in different cities scientifically and to clarify the development of different links in each city, this paper suggests an extensive assessment framework integrating rough set theory and projection pursuit-based computation to systematically assess and thoroughly evaluate the level of competitiveness of the ICV industry. First, through big data text analysis technology, we constructed a "5 + 24" two-tier evaluation index system composed of 24 level-II evaluation indexes as well as five level-I evaluation indexes and selected 19 typical cities as input data for the comprehensive evaluation system. Further, the Adaptive Random Forest based Crossover Tactical Unit (ARF-CTU) algorithm is proposed for evaluating the performance of the industrial vehicle industry. However, the ARF algorithm is employed to estimate the lowering of overfitting issues and handling of high dimensional data. Moreover, the continuously varying conditions are analyzed by CTU. Then, we constructed a comprehensive evaluation system in the rough set theory and projection pursuit: (I) Quoting the rough set non-decision-making algorithm for attribute reduction, that is, under the premise of unchanged classification ability, derive a new evaluation system, and calculate the index weight and score based on the new system. (II) Based on the projection pursuit technology, the index score is mapped by a genetic algorithm to a linear structure, and a one-dimensional projection vector is an output.

The susceptibility of mechanomyography (MMG) signals acquisition to sensor donning and doffing, and the apparent time-varying characteristics of biomedical signals collected over different periods, inevitably lead to a reduction in model recognition accuracy. To investigate the adverse effects on the recognition results of hand actions, a 12-day cross-time MMG data collection experiment with eight subjects was conducted by an armband, then a novel MMG-based hand action recognition framework with densely connected convolutional networks (DenseNet) was proposed. In this study, data from 10 days were selected as a training subset, and the remaining data from another 2 days were used as a test set to evaluate the model’s performance. As the number of days in the training set increases, the recognition accuracy increases and becomes more stable, peaking when the training set includes 10 days and achieving an average recognition rate of 99.57% (± 0.37%). In addition, part of the training subset is extracted and recombined into a new dataset and the better classification performances of models can be achieved from the test set. The method proposed effectively mitigates the adverse effects of sensor donning and doffing on recognition results.

Brain tumors are regarded as one of the most lethal forms of cancer, primarily due to their heterogeneity and low survival rates. To tackle the challenge posed by brain tumor diagnostic models, which typically require extensive data for training and are often confined to a single dataset, we propose a diagnostic model based on the Prewitt operator and a graph isomorphic network. Firstly, during the graph construction stage, edge information is extracted from MRI (magnetic resonance imaging) images using the Prewitt filtering algorithm. Pixel points with a gray value intensity greater than 128 are designated as graph nodes, while the remaining pixel points are treated as edges of the graph. Secondly, the graph data is inputted into the GIN model for training, with model parameters optimized to enhance performance. Compared with existing work using small sample sizes, the GraphMriNet model has achieved classification accuracies of 100%, 100%, 100%, and 99.68% on the BMIBTD, CE-MRI, BTC-MRI, and FSB open datasets, respectively. The diagnostic accuracy has improved by 0.8% to 5.3% compared to existing research. In a few-shot scenario, GraphMriNet can accurately diagnose various types of brain tumors, providing crucial clinical guidance to assist doctors in making correct medical decisions. Additionally, the source code is available at this link: https://github.com/keepgoingzhx/GraphMriNet.

Cities around the world have reopened from the lockdown caused by the COVID-19 pandemic, and more and more people are planning regional travel. Therefore, it is a practical problem to recommend suitable hotels to travelers amid the COVID-19 pandemic. However, it is also a challenging task since the criteria that affect hotel selection amid the COVID-19 pandemic may be different from those usually considered. From this perspective, a novel fuzzy analytic hierarchy process (FAHP)-fuzzy technique for order preference by similarity to ideal solution (fuzzy TOPSIS) approach is proposed in this study for hotel recommendation. The proposed methodology not only considers the criteria affecting hotel selection amid the COVID-19 pandemic, but also establishes a systematic mechanism to simultaneously improve the accuracy and efficiency of the recommendation process. The novel FAHP-fuzzy TOPSIS approach has been successfully applied to recommend suitable hotels to fifteen travelers for regional trips amid the COVID-19 pandemic.

Event cameras produce asynchronous discrete outputs due to the independent response of camera pixels to changes in brightness. The asynchronous and discrete nature of event data facilitate the tracking of prolonged feature trajectories. Nonetheless, this necessitates the adaptation of feature tracking techniques to efficiently process this type of data. In addressing this challenge, we proposed a hybrid data-driven feature tracking method that utilizes data from both event cameras and frame-based cameras to track features asynchronously. It mainly includes patch initialization, patch optimization, and patch association modules. In the patch initialization module, FAST corners are detected in frame images, providing points responsive to local brightness changes. The patch association module introduces a nearest-neighbor (NN) algorithm to filter new feature points effectively. The patch optimization module assesses optimization quality for tracking quality monitoring. We evaluate the tracking accuracy and robustness of our method using public and self-collected datasets, focusing on average tracking error and feature age. In contrast to the event-based Kanade–Lucas–Tomasi tracker method, our method decreases the average tracking error ranging from 1.3 to 29.2% and boosts the feature age ranging from 9.6 to 32.1%, while ensuring the computational efficiency improvement of 1.2–7.6%. Thus, our proposed feature tracking method utilizes the unique characteristics of event cameras and traditional cameras to deliver a robust and efficient tracking system.