Computational Resource Consumption Research Articles

Comparative analysis of methodologies for detecting extrachromosomal circular DNA

Extrachromosomal circular DNA (eccDNA) is crucial in oncogene amplification, gene transcription regulation, and intratumor heterogeneity. While various analysis pipelines and experimental methods have been developed for eccDNA identification, their detection efficiencies have not been systematically assessed. To address this, we evaluate the performance of 7 analysis pipelines using seven simulated datasets, in terms of accuracy, identity, duplication rate, and computational resource consumption. We also compare the eccDNA detection efficiency of 7 experimental methods through twenty-one real sequencing datasets. Here, we show that Circle-Map and Circle_finder (bwa-mem-samblaster) outperform the other short-read pipelines. However, Circle_finder (bwa-mem-samblaster) exhibits notable redundancy in its outcomes. CReSIL is the most effective pipeline for eccDNA detection in long-read sequencing data at depths higher than 10X. Moreover, long-read sequencing-based Circle-Seq shows superior efficiency in detecting copy number-amplified eccDNA over 10 kb in length. These results offer valuable insights for researchers in choosing the suitable methods for eccDNA research.

Fuzzy C-means clustering algorithm applied in computed tomography images of patients with intracranial hemorrhage

In recent years, intracerebral hemorrhage (ICH) has garnered significant attention as a severe cerebrovascular disorder. To enhance the accuracy of ICH detection and segmentation, this study proposed an improved fuzzy C-means (FCM) algorithm and performed a comparative analysis with both traditional FCM and advanced convolutional neural network (CNN) algorithms. Experiments conducted on the publicly available CT-ICH dataset evaluated the performance of these three algorithms in predicting ICH volume. The results demonstrated that the improved FCM algorithm offered notable improvements in computational time and resource consumption compared to the traditional FCM algorithm, while also showing enhanced accuracy. However, it still lagged behind the CNN algorithm in areas such as feature extraction, model generalization, and the ability to handle complex image structures. The study concluded with a discussion of potential directions for further optimizing the FCM algorithm, aiming to bridge the performance gap with CNN algorithms and provide a reference for future research in medical image processing.

Optimizing Flow and Heat Transfer With Guiding Pin Fins in a Wedge-Shaped Cooling Channel for Gas Turbine Blade Trailing Edge

Abstract In this study, an innovative guiding pin-fin array optimization has been developed through three-dimensional numerical simulations to enhance the cooling efficiency of gas turbine blade trailing edge. The guiding pin-fin array is optimized using the Kriging surrogate model and Genetic Algorithm (GA), aiming to significantly improve the heat transfer rates and uniformity within the wedge-shaped channel. The design parameter chosen for optimization is the deflection angle of each guiding pin fin, and the optimization process is conducted in two rounds. The first-round optimization yields a first-optimized guiding pin-fin array, which exhibits superior overall heat transfer performance and reasonable pressure loss compared to conventional circular, oblong, and parallel pin-fin arrays. For the first-optimized guiding pin-fin (1st-OGP) channel at the Reynolds number of Re = 50,000, the total Nusselt number and pressure loss are 44.1% higher and 9.9% lower than those of the baseline circular pin-fin array (CP), respectively. An experimental validation using the transient liquid crystal (TLC) thermography method is carried out and proves the effectiveness of the optimization process. However, it is noted that the first-optimized guiding pin-fin exhibits even lower heat transfer at a low Reynolds number of Re = 10,000, particularly in the channel middle region, which is mainly due to the incapable turning flow control in the root region of the wedged channel. To address this issue and further improve the heat transfer performance at low Reynolds numbers, a second-round optimization is performed by specifically adjusting the deflection angle of the selected guiding pin fins near the root region of the wedged channel. This secondary optimization demonstrates significant heat transfer improvements over the whole studied Reynolds number range with a reasonably reduced consumption of computational resources. The total Nusselt number and pressure loss are 69.3% higher and 11.9% lower than those of the baseline circular pin-fin array, respectively, at Re = 50,000. The optimization process proposed in this paper produces a high-performance cooling structure design with elaborate guiding pin-fin arrangements in the wedge-shaped channel, which indicates high heat transfer enhancement and relatively lower pressure loss in the wedged channel for the turbine blade trailing edge.

3D LVCN: A Lightweight Volumetric ConvNet

ABSTRACTIn recent years, with the significant increase in the volume of three‐dimensional medical image data, three‐dimensional medical models have emerged. However, existing methods often require a large number of model parameters to deal with complex medical datasets, leading to high model complexity and significant consumption of computational resources. In order to address these issues, this paper proposes a 3D Lightweight Volume Convolutional Neural Network (3D LVCN), aiming to achieve efficient and accurate volume segmentation. This network architecture combines the design principles of convolutional neural network modules and hierarchical transformers, using large convolutional kernels as the basic framework for feature extraction, while introducing 1 × 1 × 1 convolutional kernels for deep convolution. This improvement not only enhances the computational efficiency of the model but also improves its generalization ability. The pro‐posed model is tested on three challenging public datasets, namely spleen, liver, and lung, from the medical segmentation decathlon. Experimental results show that the proposed model performance has in‐creased from 0.8315 to 0.8673, with a reduction in parameters of approximately 5%. This indicates that compared to currently advanced model structures, our proposed model architecture exhibits significant advantages in segmentation performance.

An effective method for anomaly detection in industrial Internet of Things using XGBoost and LSTM

In recent years, with the application of Internet of Things (IoT) and cloud technology in smart industrialization, Industrial Internet of Things (IIoT) has become an emerging hot topic. The increasing amount of data and device numbers in IIoT poses significant challenges to its security issues, making anomaly detection particularly important. Existing methods for anomaly detection in the IIoT often fall short when dealing with data imbalance, and the huge amount of IIoT data makes feature selection challenging and computationally intensive. In this paper, we propose an optimal deep learning model for anomaly detection in IIoT. Firstly, by setting different thresholds of eXtreme Gradient Boosting (XGBoost) for feature selection, features with importance above the given threshold are retained, while those below are ignored. Different thresholds yield different numbers of features. This approach not only secures effective features but also reduces the feature dimensionality, thereby decreasing the consumption of computational resources. Secondly, an optimized loss function is designed to study its impact on model performance in terms of handling imbalanced data, highly similar categories, and model training. We select the optimal threshold and loss function, which are part of our optimal model, by comparing metrics such as accuracy, precision, recall, False Alarm Rate (FAR), Area Under the Receiver Operating Characteristic Curve (AUC-ROC), and Area Under the Precision–Recall Curve (AUC-PR) values. Finally, combining the optimal threshold and loss function, we propose a model named MIX_LSTM for anomaly detection in IIoT. Experiments are conducted using the UNSW-NB15 and NSL-KDD datasets. The proposed MIX_LSTM model can achieve 0.084 FAR, 0.984 AUC-ROC, and 0.988 AUC-PR values in the binary anomaly detection experiment on the UNSW-NB15 dataset. In the NSL-KDD dataset, it can achieve 0.028 FAR, 0.967 AUC-ROC, and 0.962 AUC-PR values. By comparing the evaluation indicators, the model shows good performance in detecting abnormal attacks in the Industrial Internet of Things compared with traditional deep learning models, machine learning models and existing technologies.

Lightweight rail surface defect detection algorithm based on an improved YOLOv8

Existing deep learning methods for rail surface defect detection face issues such as poor compatibility with embedded detection systems, high computational resource consumption, and slow detection speeds. To address these challenges, we propose a lightweight rail surface defect detection algorithm11Code: https://github.com/haichao67/GD-YOLOv8. based on an improved YOLOv8.Inspired by Huawei’s Gold-YOLO, our algorithm redesigns the neck architecture, integrates feature extraction, fusion, and injection modules, and replaces Complete Intersection over Union with Scylla Intersection over Union. Comparative experiments confirm its efficacy, achieving 94.9% precision and 81.9% recall on the Rail Surface Defect Dataset, a 26.3% increase in frames per second, and a model size reduction to 63%. Our algorithm not only ensures satisfactory detection results but also reduces the number of model parameters and increases the detection speed.

Leveraging Incremental Learning for Dynamic Modulation Recognition

Modulation recognition is an important technology used to correctly identify the modulation modes of wireless signals and is widely used in cooperative and confrontational scenarios. Traditional modulation-recognition algorithms require the assistance of expert experiences, which constrains their applications. With the rapid development of artificial intelligence in recent years, deep learning (DL) is widely advocated for intelligent modulation recognition. Typically, DL-based modulation-recognition algorithms implicitly assume a relatively static scenario in which the signal samples of all the modulation modes can be completely collected in advance. In practical situations, the radio environment is quite dynamic and the signal samples with new modulation modes may appear sequentially, in which the current DL-based modulation-recognition algorithms may require unacceptable time and computing resource consumption to re-train the optimal model from scratch. In this study, we leveraged incremental learning (IL) and designed a novel IL-based modulation-recognition algorithm that consists of an initial stage and multiple incremental stages. The main novelty of the proposed algorithm lies in the new loss function design in each incremental stage, which combines the distillation loss of recognizing old modulation modes and the cross-entropy loss of recognizing new modulation modes. With the proposed algorithm, the knowledge of the signal samples with new modulation modes can be efficiently learned in the current stage without forgetting the knowledge learned in the previous stages. The simulation results demonstrate that the proposed algorithm could achieve a recognition accuracy close to the upper bound with a much lower computing time and it outperformed the existing IL-based benchmarks.

A boundary-type algorithm based on the discrete ordinates for neutron transport equation

The spatial distribution of neutron flux in the core of a nuclear reactor plays a crucial role in ensuring nuclear safety. It can be determined by solving the neutron transport equation, where computational resource consumption is gradually receiving more attention, especially in problems with multiple reflective boundary conditions. It is common practice to assume initial boundary values in one angular direction to obtain a global solution, and then use reflective boundary conditions to solve for the other directions, iterating until convergence. However, this approach has the drawback of requiring the global solution, and each iteration necessitates storing the neutron flux values, consuming time and storage space. In order to address this issue, this paper proposes a precise and efficient boundary-type method, the Half Boundary Method (HBM). This method establishes relationships between adjacent node values through integration and mathematical deduction, ultimately obtaining the relationship among any node and the boundary values in any direction. As a result, iteration is only required for boundary values, reducing the iteration amount and thus saving time and storage space. Using the discrete ordinate (SN) method for angular discretization and HBM for spatial discretization, this paper solved the steady-state neutron transport equation for two-dimensional systems modeled with Cartesian geometry. The method is validated using several benchmark problems, including the 2D ISSA benchmark, the 2D mono-group k-eigenvalue problem and BWR rod bundle test problem. All of these problems demonstrate good results and effectively reduced computational time.

Fault diagnosis for multiple redundancy aileron actuator based on parallel-SDP and polar sparse representation

Abstract As a critical component of aircraft flight control systems, the aileron actuator's fault directly affects the performance of flight control performance system and the overall flight safety of aircraft. Therefore, effective fault diagnosis of the aileron actuator becomes particularly important. However, with the development of redundancy design, the structure of aileron actuators more and more complex, and the fault modes are more and more diverse, which increases the difficulty of fault diagnosis significantly. To address this challenge, this study proposes a fault diagnosis method based on Parallel-SDP and Polar Sparse Representation.
First, the current residual signals of force motors are obtained using bi-step observer. Then, the residual signals are transformed into Parallel-SDP images, where each of the four petals of Parallel-SDP image generated from the corresponding channel. The Parallel-SDP image presents global fault information and capture subtle changes of residual signals, which increases the sensitivity of fault diagnosis. Subsequently, rapid single-channel fault localization is achieved by barycenter analysis of Parallel-SDP images. Finally, based on the result of fault localization, the proposed polar sparse representation algorithm is utilized for fault diagnosis of mechanical and control faults separately. Based on the dataset obtained from physical-parameter-simulation, the proposed method was validated, and the accuracy of fault diagnosis of was 99.29%, which is better than conventional fault diagnosis, meanwhile, the computational resources consumption of the proposed method is much less than deeplearning-based method, which is suitable for embedded computing system in aircraft.

A lightweight visual mamba network for image recognition under resource-limited environments

The Vision Transformers (ViTs) models show great potential in recognition due to their excellent self-attention mechanisms. However, they often need more computational complexity, making achieving high performance in resource-constrained environments challenging. This paper proposes a lightweight visual model (called LightViM) to tackle these challenges, specifically devised for recognition challenges under resource-constrained environments. Considering the imperative of reducing the model’s computational complexity and resource consumption , we propose lightweight local–global feature fusion modules based on Mamba (LGF-Mamba) aimed at integrating spatially detailed local information with globally contextualized features while maintaining lower linear time complexity. In LGF-Mamba, Mamba sub-blocks are first designed to extract global information; then, for the rapid extraction of local features, LocalE module is designed and integrated into LGF-Mamba, exploring more comprehensive feature representation. This mechanism efficiently directs the network to integrate features from different spatial scales, thereby improving the recognition performance in resource-constrained environments. Experimental results indicate that, in comparison to other leading-edge lightweight methods, the proposed procedure simultaneously achieves superior recognition performance and minimal resource consumption.

Current progress in subseasonal-to-decadal prediction based on machine learning

The application of machine learning (ML) techniques to climate science has received significant attention, particularly in the field of climate predictions, ranging from sub-seasonal to decadal time scales. This paper reviews recent progress of ML techniques employed in climate phenomena prediction and the enhancement of dynamic forecast models, which provide valuable insights into the great potentials of ML techniques to improve climate prediction capabilities with reduced computational time and resource consumption. This paper also discusses several major challenges in the application of ML to climate prediction, including the scarcity of datasets, physical inconsistency, and lack of model transparency and interpretability. Additionally, this paper sheds light on how climate change impacts ML model training and prediction, and explores three key areas with potential breakthroughs: large-scale climate models, knowledge discovery driven by ML, and hybrid dynamical-statistical models, underscoring the important role of the integration of “ML and dynamical models” in building a bridge between the artificial intelligence and climate science.

Three-dimensional particulate volume fraction reconstruction in the fluid based on the Lambert–Beer physics information neural network

The measurement of particle volume fraction in flow fields is of great significance in scientific research and engineering applications. As one of the particle detection techniques, the light extinction method is widely used in measuring nano-particles volume fraction in flow fields due to its simplicity and non-contact nature. In particular, in complex reactive flow fields like combustion reactions, the volume fraction of soot particulate and other particles can be accurately measured and reconstructed via the light extinction method that based on the Beer–Lambert law. This is crucial for exploring combustion phenomena, understanding their internal mechanisms, and reducing pollutant emissions. However, due to the enormous computational burden, current algebra reconstruction techniques struggle to achieve high-precision three-dimensional (3D) reconstruction of particles. Therefore, this paper originally proposes a 3D reconstruction algorithm based on the Beer–Lambert law physical information neural networks (LB-PINNs). By incorporating physical information as constraints into the particle reconstruction process, it is possible to achieve high-precision 3D reconstruction of particles in complex flow field environments with low computational cost. Meanwhile, to address the trade-off issues of reconstruction accuracy and smooth noise resistance in previous reconstruction algorithms, i.e., Tikhonov regularization, this paper employs dynamically adjusted regularization parameters in the LB-PINN algorithm. This approach ensures smooth noise-resistant processing while maintaining reconstruction accuracy, significantly reducing computation time and resource consumption. According to the experimental results, LB-PINNs demonstrate superior performance compared to previous reconstruction algorithms when reconstructing the soot volume fraction in complex reacting flow fields, i.e., combustion flame scenarios.

Image compression encryption algorithm combining two-dimensional modular hyperchaotic map and compressed sensing

Recently, to offer better ensure for image privacy security, numerous new image encryption algorithms have been proposed. However, these algorithms still suffer from the problems of chaotic performance scarcity, low encryption effect, and high consumption of computational resources. To solve the above issues, we first construct a two-dimensional modular hyperchaotic map (2D-MHM). Then, we further develop an image encryption algorithm based on 2D-MHM and compressed sensing (CS). Several chaotic metrics verify the randomness and validity of 2D-MHM. These metrics include bifurcation diagram, Lyapunov exponent, initial value sensitivity, 0–1 test, and NIST test. Specifically, CS significantly reduces the ciphertext image size thereby reducing its resource consumption during transmission. Reality-preserving fractional DCT (RP-Fdct) diffusion is utilized to transform pixels into the frequency domain to enhance the encryption effect. Subsequently, lightweight index confusion and XOR diffusion further improve the algorithm security. The security of the algorithm is verified through various experiments. It is able to encrypt grayscale and color images of different sizes with good results. Notably, this algorithm also implements the encryption requirements for binary images. Due to our designs, it outperforms recently reported encryption algorithms in several areas, especially in reconstruction performance.

High-Performance Grape Disease Detection Method Using Multimodal Data and Parallel Activation Functions

This paper introduces a novel deep learning model for grape disease detection that integrates multimodal data and parallel heterogeneous activation functions, significantly enhancing detection accuracy and robustness. Through experiments, the model demonstrated excellent performance in grape disease detection, achieving an accuracy of 91%, a precision of 93%, a recall of 90%, a mean average precision (mAP) of 91%, and 56 frames per second (FPS), outperforming traditional deep learning models such as YOLOv3, YOLOv5, DEtection TRansformer (DETR), TinySegformer, and Tranvolution-GAN. To meet the demands of rapid on-site detection, this study also developed a lightweight model for mobile devices, successfully deployed on the iPhone 15. Techniques such as structural pruning, quantization, and depthwise separable convolution were used to significantly reduce the model’s computational complexity and resource consumption, ensuring efficient operation and real-time performance. These achievements not only advance the development of smart agricultural technologies but also provide new technical solutions and practical tools for disease detection.

Short-term air quality prediction using point and interval deep learning systems coupled with multi-factor decomposition and data-driven tree compression

Clean air, as a symbol of high-quality air quality, is the most basic requirement for people to maintain health. Moreover, in keeping humans fit, accurate short-term air quality prediction is vital. The decomposition algorithm can better capture the local features and temporal changes of the data. However, it increases the computation time, resource consumption, and complexity of the model. On the other hand, existing forecasting systems overlook instability and uncertainty. To solve the above problems, a deterministic and uncertainty AOA-DBGRU-MDN deep learning systems is proposed, which combines arithmetic optimization algorithm (AOA), double-layer bi-directional GRUs (DBGRU), and mixture density network (MDN). The above systems consider meteorological factors and air pollutants comprehensively. It involves feature selection using maximum information coefficient (MIC), decomposition using complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) algorithm, classification, and compression of decomposed components using entropy-Huffman tree compression. Firstly, the information measurement process reduces the number of components significantly. Following the incorporation of multi-factor data, the optimal DBGRU model is then obtained using AOA. Finally, the training errors are fitted using MDN to obtain interval prediction results. The experiments demonstrate that (1) Using the CEEMDAN algorithm can improve the prediction accuracy; (2) Classifying and reconstructing the data based on entropy-Huffman tree compression can not only decrease the model's training volume and improve training efficiency but also boost the model's prediction accuracy; (3) The AOA-DBGRU-MDN system performs probabilistic prediction to obtain an effective and intuitive prediction interval to improve the point prediction of air quality prediction.

Research on Space Operation Control of Air Float Satellite Simulator Based on Constraints Aware Particle Filtering-Nonlinear Model Predictive Control

This paper addresses the challenges of close proximity operations, such as rendezvous, docking, and fly-around maneuvers for micro/nano satellites, which require high control precision under the low power and limited computational capabilities of spacecraft. Firstly, a three-degree-of-freedom air float simulator platform is designed for ground-based experiments. Subsequently, model predictive controllers based on constraints aware of particle filtering (CAPF-NMPC) are developed for executing operations such as approach, fly-around, and docking maneuvers. The results validate the effectiveness of the experimental system, demonstrating position control accuracy less than 0.03 m and attitude control accuracy less than 3°, maintaining lower computational resource consumption. This study offers a practical solution for the onboard deployment of optimized control algorithms, highlighting significant value for further engineering applications.

Global feature capture and spatially-aware neural networks for predicting CO2-flooding performance in heterogeneous low-permeable reservoirs

CO2-flooding can enhance crude oil recovery rates by reducing the viscosity of crude oil, improving the mobility ratio between crude oil and water, and leveraging the swelling effects of crude oil components. Furthermore, it facilitates the sequestration of greenhouse gases, leading to the realization of carbon capture, utilization, and storage (CCUS), which holds significant strategic importance for mitigating climate change, safeguarding ecological environments, and achieving sustainable development. Consequently, predicting CO2-flooding performance in heterogeneous low-permeability reservoirs holds crucial significance in sustainable energy production. However, the drawbacks of numerical simulation techniques, such as extensive computational resource consumption and prolonged simulation durations, have prompted extensive research into the utilization of artificial intelligence technologies. Prevalent approaches often treat the problem as a time series issue, which proves inadequate for reservoirs lacking historical production data. Furthermore, this methodology is unable to fully exploit the complex implicit relationships between various factors and the prediction target. Our study proposes a novel method for predicting the performance, which stands apart from previous approaches by not treating the problem as a time series issue, eliminating the need for historical production data for each prediction, and can predict various production variables such as gas production, oil production, water production, etc. in a universal manner. We combine enhanced MultiScale Non-local Neural Networks with ConvNet to handle sparse well network matrices with injection-production parameters. Additionally, we employ MultiScale convolutional networks with spatial attention mechanism to address formation permeability fields with spatial distribution features. Experimental results demonstrate the superiority of our approach in accuracy, achieving a prediction accuracy of over 97%. This research provides an innovative and flexible method, making significant contributions to the intelligentization and sustainable development of the petroleum industry, aligned with global efforts to both reduce carbon emissions and neutralize the existing ones.

Event-triggered lifted robust MPC stabilization control for space telescope on dual-rate actuated rigid spacecraft systems

This study addresses the pose stabilization control problem for enhancing a space telescope mounted on a spacecraft system under a dual-rate actuated setup. An input-lifting approach is utilized to manage the dual rates in the control components, specifically the orbital space telescope and its loading platform. To counteract external spatial perturbations and spacecraft system uncertainties, an integrated control scheme is proposed, combining real-time model parameter estimation, active compensation control, and event-triggered lifted robust model predictive control (ET-LRMPC). The event-triggered mechanism in the proposed algorithm minimizes computational resource consumption while maintaining effective spacecraft control. Numerical simulations demonstrate that the proposed control scheme achieves favorable results under persistent external perturbations and model uncertainties. In terms of the overshoot, the algorithm proposed reduces the control effect by up to 87.1% in the spatial displacement of the payload and 98.6% in the Euler attitude angle compared to the conventional control algorithm. For the control of the base, the amount of overshoots with respect to the conventional control in spatial displacement and Euler attitude angle is reduced by 49% and 97.1%.

A deep transfer learning approach for cross-subject recognition of mental tasks based on functional near-infrared spectroscopy

In the field of brain-computer interfaces (BCIs) based on functional near-infrared spectroscopy (fNIRS), traditional subject-specific decoding methods suffer from the limitations of long calibration time and low cross-subject generalizability, which restricts the promotion and application of BCI systems in daily life and clinic. To address the above dilemma, this study proposes a novel deep transfer learning approach that combines the revised inception-residual network (rIRN) model and the model-based transfer learning (TL) strategy, referred to as TL-rIRN. This study performed cross-subject recognition experiments on mental arithmetic (MA) and mental singing (MS) tasks to validate the effectiveness and superiority of the TL-rIRN approach. The results show that the TL-rIRN significantly shortens the calibration time, reduces the training time of the target model and the consumption of computational resources, and dramatically enhances the cross-subject decoding performance compared to subject-specific decoding methods and other deep transfer learning methods. To sum up, this study provides a basis for the selection of cross-subject, cross-task, and real-time decoding algorithms for fNIRS-BCI systems, which has potential applications in constructing a convenient and universal BCI system.

MolCFL: A personalized and privacy-preserving drug discovery framework based on generative clustered federated learning

In today’s era of rapid development of large models, the traditional drug development process is undergoing a profound transformation. The vast demand for data and consumption of computational resources are making independent drug discovery increasingly difficult. By integrating federated learning technology into the drug discovery field, we have found a solution that both protects privacy and shares computational power. However, the differences in data held by various pharmaceutical institutions and the diversity in drug design objectives have exacerbated the issue of data heterogeneity, making traditional federated learning consensus models unable to meet the personalized needs of all parties. In this study, we introduce and evaluate an innovative drug discovery framework, MolCFL, which utilizes a multi-layer perceptron (MLP) as the generator and a graph convolutional network (GCN) as the discriminator in a generative adversarial network (GAN). By learning the graph structure of molecules, it generates new molecules in a highly personalized manner and then optimizes the learning process by clustering federated learning, grouping compound data with high similarity. MolCFL not only enhances the model’s ability to protect privacy but also significantly improves the efficiency and personalization of molecular design. MolCFL exhibits superior performance when handling non-independently and identically distributed data compared to traditional models. Experimental results show that the framework demonstrates outstanding performance on two benchmark datasets, with the generated new molecules achieving over 90% in Uniqueness and close to 100% in Novelty. MolCFL not only improves the quality and efficiency of drug molecule design but also, through its highly customized clustered federated learning environment, promotes collaboration and specialization in the drug discovery process while ensuring data privacy. These features make MolCFL a powerful tool suitable for addressing the various challenges faced in the modern drug research and development field.