Fast Optimization Research Articles

The human iris contains rich physiological information through its texture, color, and morphology. Traditional iris analysis methods are limited by computational inefficiency and poor scalability for multi-indicator prediction. This study aimed to develop and validate a lightweight, multi-task deep learning framework that can noninvasively predict multiple physiological indicators from high-resolution iris images. We developed a hybrid model combining ConvNeXt-Tiny and TinyViT backbones, trained on 30,353 iris images with a multi-task learning (MTL) strategy. To enhance performance with limited labeled data, semi-supervised learning with consistency regularization and pseudo-labeling was integrated. Fast Adaptive Multi-Task Optimization (FAMO) and Nash Multi-Task Learning (Nash-MTL) were used to balance task-specific losses. Feature vectors with 2400 dimensions — derived from texture, chromatic, and morphological cues — were compressed to 120 dimensions using Principal Component Analysis (PCA), preserving 96.2% of the data variance. This dimensionality reduction led to a 2.9-fold improvement in inference speed and reduced feature storage size by nearly 80% significantly enhancing computational efficiency without compromising predictive accuracy. The model achieved 89.1% average accuracy across 13 physiological tasks, with F1 scores of 0.91 for heart and 0.89 for brain indicators, outperforming fully supervised baselines. Among these, 7 organ-related indicators were validated using data from 60 Parkinson’s disease patients, showing strong left–right eye correlation ([Formula: see text], [Formula: see text]).

To fully exploit the truck’s energy-saving potential, optimization-based methods are typically used to control the powertrain system, in which Predictive Cruise Control (PCC) is an effective method for improving energy efficiency. Due to nonlinear characteristics and long prediction horizon, the optimization problem becomes too complex to solve, making it difficult to meet the real-time computing requirements of embedded on-board platforms. This paper proposes a fast solution algorithm for PCC based on Pontryagin’s Maximum Principle (PMP), where warm-start mechanism is applied to accelerate the convergence speed of the solution searching bisection method, resulting in a 70.83% improvement in calculation speed compared to the standard bisection method. Finally, the proposed algorithm is tested on commercially heavy-duty trucks. Experimental results show that in high-speed conditions, fuel economy is improved by 4.66% over the rule-based PCC system. Additionally, the algorithm demonstrates improvements of over 5% in driving comfort and over 6% in speed tracking performance.

To integrate Attribute-Based Encryption (ABE) and Fully Homomorphic Encryption (FHE) within the NS-2 simulation environment, we propose a novel simulation model called FHE and ABE with Fast Exponentiation Optimization (FA-FEO) for smart city environment monitoring. This model evaluates important performance metrics like throughput, latency, memory utilization, power consumption, etc. With networked sensors and devices, the IoT enables efficient data collection and monitoring, but challenges like safe data transfer with energy constraints, and privacy preservation remain crucial. To provide strong data security and privacy while permitting smooth communication across decentralized IoT networks, our approach (FA-FEO) places a strong emphasis on the employment of FHE and ABE. A study of the performance of IoT network communication under basic implementation of FHE, two types of ABE like Ciphertext-policy ABE (CP-ABE) and Key-policy ABE (KP-ABE), and a BASE encryption indicates the significance of both ABE and FHE for practical smart city applications. The proposed model has been evaluated in detail using simulations for smart city environment monitoring scenarios and the results show that it is possible to deal with the overhead caused by FHE and ABE, guaranteeing safe and efficient energy-efficient solutions for scenarios such as environmental monitoring in smart cities.

FTO-SORT: a fast track-id optimizer for enhanced multi-object tracking with SORT in unseen pig farm environments

A fundamental challenge in lithium-ion battery charging is the inherent trade–off between charging speed and battery health. Fast charging tends to accelerate battery degradation, while slow charging extends downtime and intensifies range anxiety, heightening concerns over inadequate driving range during operation. This contradiction has become a key bottleneck restricting the advancement of electric vehicles. In response to the limitations of conventional charging strategies and optimization methods, which typically intensify this trade–off, this study proposes a novel two–stage fast charging optimization strategy for lithium–ion batteries. The proposed method first introduces a hybrid clustering algorithm that combines the canopy algorithm with bisecting K–means to achieve adaptive SOC staging. This staging is guided by the nonlinear characteristics of the internal resistance with respect to the state of charge (SOC), allowing for a data–driven division of charging phases. Following staging, a closed–loop optimization framework is developed. A wavelet neural network (WNN) is employed to precisely capture and approximate the nonlinear characteristics of the charging process for performance prediction, upon which a multi–strategy enhanced multi–objective particle swarm optimization (MOPSO) algorithm is applied to efficiently search for Pareto–optimal solutions that balance charging time and ohmic loss. In addition, an active learning mechanism is incorporated to refine the WNN using selectively sampled data iteratively, thereby improving prediction accuracy and the robustness of the optimization process. Experimental results demonstrate that when the SOC reaches 70%, the proposed method shortens the charging time by 12.5% and reduces ohmic loss by 31% compared with the conventional constant current–constant voltage (CC–CV) strategy, effectively achieving a balance between charging efficiency and battery health.

Automatic differentiation (AD), a key method for accurately and efficiently computing derivatives in modern machine learning, is now implemented in Phoenix®NLME™8.6 for the first time and applied to the first-order conditional estimation extended least squares (FOCE ELS), Laplacian, and adaptive Gaussian quadrature (AGQ) algorithms. We name the AD implementation as 'automatic-differentiation-assisted parametric optimization' (ADPO), which can be enabled by checking the 'Fast Optimization' option. We present in detail how ADPO is implemented in the frequently used FOCE ELS algorithm, and analyze its performance from the benchmarks based on four PK/PD models. We show both ADPO and traditional FOCE ELS which uses gradients obtained from finite difference (FD) are reasonably accurate and robust, while the main advantage of ADPO being that it considerably reduces computation time no matter what ODE solvers are used: in general ADPO reduces the total run time by around 20% to 50% compared to traditional FOCE ELS. In a case for the realistic voriconazole model using 'auto-detect' ODE solver, 95% reduction in the total run time is observed.

BackgroundMultichannel transcranial magnetic stimulation (mTMS) enables electronic steering of induced electric fields across multiple cortical targets without physical coil repositioning, addressing key limitations of conventional single-channel TMS (sTMS). However, determining optimal input currents for focal stimulation remains challenging, and different mTMS systems have not been systematically compared under realistic hardware constraints.ObjectiveTo develop an user-centric framework for optimizing and assessing mTMS focality by introducing a generic optimization algorithm, establishing meaningful focality metrics, and comparing mTMS coil arrays with traditional single-channel TMS across cortical targets.MethodsWe developed a fast optimization framework incorporating target E-field constraints via parametrization of degenerated hyperellipsoids, explicitly integrating current-rate limits, for example from stimulator electronics and coil heating. Using high-resolution finite-element models of nine individual brains, we compared two mTMS designs (5-channel planar and 6/12-channel spherical systems) with standard sTMS figure-of-eight coils. Three complementary metrics quantified performance: Focality, Target2Max, and OverstimulatedArea.ResultsDespite using a single optimized placement for all region-of-interest targets, mTMS achieved focality comparable to repositioned single-channel TMS. For superficial targets, single-channel TMS showed slightly better focality, but for deeper cortical targets (>25mm skin-cortex distance), mTMS performed similarly. More stimulation channels improved focality but required stronger current-rate constraints. The planar design performed better for deeper targets, while spherical designs improved with additional channels.ConclusionmTMS systems demonstrate remarkable performance comparable to standard TMS, enabling efficient multi-target stimulation without repositioning. Our open-source framework provides practical tools for designing and evaluating mTMS systems, supporting goal-directed mTMS development and effective application.

Purpose:Passive resonators have been widely used in MRI to manipulate RF field distributions. However, optimizing these structures using full-wave electromagnetic (EM) simulations is computationally prohibitive, particularly for massive-element passive resonator arrays with many degrees of freedom.Methods:While the EM and RF circuit co-simulation method has previously been applied to RF coil design, this work presents, for the first time, a co-simulation framework tailored specifically for the analysis and optimization of passive resonators. The framework performs a single full-wave EM simulation in which the resonator’s lumped components are replaced by ports, followed by circuit-level computations to evaluate arbitrary capacitor/inductor configurations. This allows integration with a genetic algorithm to rapidly optimize the resonator parameters to enhance fields in a targeted region of interest (ROI).Results:The proposed method was validated across three scenarios of increasing complexity: (1) a single-loop passive resonator on a spherical phantom, (2) a two-loop array on a cylindrical phantom, and (3) a two-loop array on a human head model. In all cases, the co-simulation results showed excellent agreement with full-wave EM simulations, with relative errors below 1%. The genetic-algorithm-driven optimization, involving tens of thousands of capacitor combinations, completed in under 5 minutes—whereas equivalent full-wave EM sweeps would require an impractically long computation time.Conclusion:This work extends co-simulation methodology to passive resonator design for the first time, enabling fast, accurate, and scalable optimization. The approach significantly reduces computational burden while preserving full-wave accuracy, making it a powerful tool for passive RF structure development in MRI.

Purpose This study/paper aims to introduces a fast and efficient stochastic chaos game optimization (CGO) algorithm for pattern synthesis of linear antenna array (LAA). Design/methodology/approach CGO algorithm is used to determine the optimal current amplitude and the interelement distance between the array elements of an LAA. Eleven design examples are simulated to explore the ability of CGO algorithm in minimizing peak and close-in side lobe levels, null placement and side lobe power reduction under the constraint of specified beamwidth. Findings The simulation results suggest that CGO algorithm outperforms other nature-inspired metaheuristic optimization algorithms. The fast convergence nature of CGO algorithm for discovering the optimal amplitude and position shows its usefulness for real-time applications. Originality/value CGO algorithm is first time explored for pattern synthesis of LAA. To analyze its effectiveness, two design examples are simulated in electromagnetic (EM) simulator and the result shows that MATLAB and EM simulator findings are really closely aligned. The results are also validated by the Wilcoxon and Friedman statistical test.

To mitigate the challenges of inaccuracies and inefficiencies in manual alignment of hydroturbine shafts, a wireless measuring instrument for axial alignment based on extremum fast optimization is designed. The proposed method introduces three primary innovations. First, a wireless sensing and measurement system is established by strategically deploying grating displacement sensors and angular encoders at multiple critical locations, integrated with LoRa wireless transmission technology. This configuration enables 360° continuous data acquisition, effectively eliminating the constraints of traditional eight-point discrete measurement methods that require fixed-point positioning and manual coordination. Second, an enhanced simulated annealing (SA) extremum optimization algorithm is developed, incorporating an adaptive step-size perturbation model, a multi-stage annealing process with a restart mechanism, and a flexible temperature decay strategy. These improvements address the common shortcomings of conventional SA algorithms, such as fixed step sizes that cause local optimum trapping and slow convergence, thereby enabling precise and rapid identification of extremum points in axis deflection curves. Third, sinusoidal function fitting is employed for deflection curve analysis, which, when combined with the optimized algorithm, allows direct determination of the maximum deviation direction, replacing the traditional multi-iteration adjustment process. The axial alignment experiments are conducted on the hydroturbine axis prototype, and the results show that the swing values obtained from the designed measuring instrument are consistent with the traditional eight-point method, achieving 99.8% consistency and improving alignment efficiency by 400%. The extremum identification accuracy reaches 0.001mm, demonstrating that the proposed method establishes a novel and highly effective paradigm for intelligent hydroturbine shaft alignment.

Dynamic electricity-carbon factors driven fast day-ahead coordination and optimization of source-load-storage in distribution networks with a white box accelerator

Meta-tuning and fast optimization of machine learning models for dynamic methane prediction in anaerobic digestion.

To address the impact of distributed energy resource volatility on distribution network fault restoration, this paper proposes a strategy that incorporates net load forecasting. A Bayesian-optimized long short-term memory neural network is used to accurately predict the net load within fault-affected areas, achieving an R2 of 0.9569 and an RMSE of 12.15 kW. Based on the forecasting results, a fast restoration optimization model is established, with objectives to maximize critical load recovery, minimize switching operations, and reduce network losses. The model is solved using a genetic algorithm enhanced with quantum particle swarm optimization (GA-QPSO), a hybrid metaheuristic known for its superior global exploration and local refinement capabilities. GA-QPSO has been successfully applied in various power system optimization problems, including service restoration, network reconfiguration, and distributed generation planning, owing to its effectiveness in navigating large, complex solution spaces. Simulation results on the IEEE 33-bus system show that the proposed method reduces network losses by 33.2%, extends the power supply duration from 60 to 120 min, and improves load recovery from 72.7% to 75.8%, demonstrating enhanced accuracy and efficiency of the restoration process.

Given that the decision tree C4.5 algorithm has outstanding performance in prediction accuracy on medical datasets and is highly interpretable, this paper carries out an optimization study on the selection of hyperparameters of the algorithm in order to achieve fast and accurate optimization of the algorithm model. The decision tree models are first constructed by taking different values of hyperparameters, and then the performance of each model is evaluated, and then the evaluated data are associated and integrated with the character metadata of the dataset. Three evaluation values of accuracy, AUC and F1-measure and 293 basic data sets were used to build a meta-database of hyperparameter M optimization required by the study. And then the range of values of C4.5 algorithm hyperparameters corresponding to the different character datasets are recommended through the modeling learning. The results show that for more than 65% of the data sets, there is no need to tune the hyperparameter M, which can avoid the waste of time caused by unnecessary tuning. The accuracy rate of the hyperparameter optimization value judgment model obtained in this study can reach more than 80%. The test and evaluation results verify the feasibility of the optimized hyperparameter value recommendation, which provides an important basis for the fast tuning and optimization of the C4.5 algorithm parameters.

Magnesium (Mg) alloys show promise for lightweight structural and biomedical applications, but they face challenges such as poor corrosion resistance and complex deformation behavior. This systematic review explores how Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL) address these limitations. These techniques enable the fast and accurate prediction and optimization of material properties, thereby reducing experimental effort and accelerating the design of high-performance Mg alloys. A multi-database validation approach using Scopus and Web of Science ensured methodological robustness when searching for AI, ML, and DL in Mg alloys. A comparative analysis of author keywords, index keywords, sources, authors, and countries confirmed strong thematic consistency between databases, thereby enhancing the credibility of the cluster-based bibliometric analysis. The PRISMA framework was used to ensure the structured literature search, eligibility assessment, and documentation of the selection process. 185 peer-reviewed articles (2015–2025) were analyzed and organized into seven refined thematic clusters: ‘mechanical behavior modeling using neural networks’, ‘AI-driven alloy design and compositional optimization’, ‘atomic-scale modeling and physics-guided learning’, ‘AI applications in welding and thermomechanical processing’, ‘biomaterials and microstructural optimization’, ‘corrosion modeling and degradation prediction’, ‘data-driven design and integrated optimization frameworks’. The review highlights the extensive application of models, including Artificial Neural Networks, Convolutional Neural Networks, and hybrid frameworks that combine ML with optimization algorithms or physical simulations. These approaches enhance predictions on mechanical properties, microstructural changes, corrosion behavior, and processing results of Mg alloys. The study also discusses cross-cutting themes such as simulation speed-up metrics, model interpretability across domains, and limitations in dataset coverage. Findings indicate AI-based methods can expedite alloy design and performance optimization; however, challenges remain in data accessibility, model interpretability, and experimental validation. The study concludes that integrating physics-informed ML models, using multimodal data, and employing inverse design will be crucial for advancing the intelligent development of high-performance Mg alloys for sustainable engineering applications.

Guidelines emphasize the importance of quadruple therapy in heart failure with reduced ejection fraction (HFrEF), and of achieving target doses quickly. The aim of the MAR-HF-Titration Study was to assess the effectiveness of a nurse-led, telemedicine program for this purpose. Prospective study including 210 HFrEF patients between September 2021 and December 2023. The 6-month follow-up used a telemedicine program involving daily telemonitoring of vital signs, weight and symptoms, routine laboratory test monitoring, videoconferences and the use of algorithms for drug titration. Mean age was 70 years, 35% women. Mean left ventricle ejection fraction (LVEF) was 31%, 60% had a non-ischemic etiology, and 61% newly diagnosed HF. After the intervention, significant increases were achieved in the use of quadruple therapy at any dose (63% vs 75%; p=0.001), at ≥50% of the target doses of each drug (10% vs 44%, p<0.001), and at target doses of all drugs (0% vs 19%, p<0.001). Median titration time was 45 days for betablockers, 65 for MRA, and 69 for ARNI. Older age, chronic kidney disease, hypotension, and higher NT-proBNP were associated with lower odds of effective titration. The LVEF increased (31% to 45%, p<0.001) and the NT-proBNP values were significantly reduced (1960 to 806 pg/ml, p<0.001). Increases in the use of prognostic medications were also observed in a sub-analysis restricted to participants aged ≥75 years. Our intensive, nurse-led, telemedicine-based program attained a fast optimization of prognostic therapies in HFrEF and resulted in significant improvements in parameters with prognostic relevance.

In this study, a high-efficiency axial fan design method and process are proposed by integrating the controlled blading design (CBD) method for the spanwise design of blade angles, the through-flow analysis method incorporating three-dimensional flow effects for performance prediction of the designed fan, and an optimization algorithm suitable for multi-variable problems. The main objective of this study is to obtain the optimal spanwise distribution of blade angles and chord length. To achieve this, the three-dimensional blade design process of the axial flow fan is established using the CBD model, in which the camber angle and setting angle along the spanwise direction are set as design variables, while the chord length along the spanwise direction is considered as another design variable. To predict the performance and efficiency of the designed fan, the through-flow analysis method is introduced, and the accuracy of flow and performance predictions using this method is verified by comparing with measurement results. A newly developed hybrid metaheuristic algorithm is applied as an optimization technique in the fan design and through-flow analysis program, enabling the optimal design of a high-efficiency axial flow fan. An optimization problem maximizing fan efficiency is defined and several design constraints are also set. The optimization algorithm is applied to the fan design and through-flow analysis program, achieving a very fast and simple optimization process and obtaining the optimal axial fan model. By comparing the optimal fan model with the initial fan model based on the free-vortex flow type, it is confirmed that fan efficiency is improved by 4.2 percentage points through this optimization. To verify the reliability of this optimization design method, CFD analysis, manufacturing, and testing are conducted for the optimized fan model. A comparison between the optimal design results and CFD calculation results demonstrates that this optimization method has very high predictive accuracy and design reliability. Furthermore, by comparing the design and CFD results of the optimized model with actual performance test results, the improvement in performance and efficiency through this optimization design method is validated. Additionally, the optimized axial fan derived in this study exhibits excellent performance characteristics, maintaining high efficiency and low power characteristics even under low flow conditions.

This study presents Fast paRticle thErapy Dose optimizer (FREDopt), a newly developed GPU-accelerated open-source optimization software for simultaneous proton dose and dose-averaged linear energy transfer (LETd) optimization in intensity-modulated proton therapy treatment planning. FREDopt was implemented entirely in Python, leveraging CuPy for GPU acceleration and incorporating fast Monte Carlo simulations from the FRED code. The treatment plan optimization workflow includes pre-optimization and optimization, the latter equipped with a novel superiorization of feasibility-seeking algorithms. Feasibility-seeking requires finding a point that satisfies prescribed constraints. Superiorization interlaces computational perturbations into iterative feasibility-seeking steps to steer them toward a superior feasible point, replacing the need for costly full-fledged constrained optimization. The method was validated on two treatment plans of patients treated in a clinical proton therapy center, with dose and LETddistributions compared before and after reoptimization. Simultaneous dose and LETdoptimization using FREDopt led to a substantial reduction of LETdand (dose) × (LETd) in organs at risk while preserving target dose conformity. Computational performance evaluation showed execution times of 14-50 min, depending on the algorithm and target volume size-satisfactory for clinical and research applications while enabling further development of the well-tested, documented open-source software.

Determining the intricate structure and function of neural circuits requires the ability to precisely manipulate circuit activity. Two-photon holographic optogenetics has emerged as a powerful tool for achieving this via flexible excitation of user-defined neural ensembles. However, the precision of two-photon optogenetics has been constrained by off-target stimulation, an effect where proximal non-target neurons can be unintentionally activated due to imperfect spatial confinement of light onto target neurons. Here, we introduce a real-time computational approach to mitigating off-target stimulation by first empirically sampling each neuron’s sensitivity to stimulation at proximal locations, and then optimizing stimulation sites using a fast, interpretable model based on adaptive non-negative basis function regression (NBFR). NBFR is highly scalable, completing model fitting for hundreds of neurons in just a few seconds and then optimizing stimulation sites in several hundred milliseconds per stimulus – fast enough for most closed-loop behavioral experiments. We characterize the performance of our approach in both simulations and in vivo experiments in mouse hippocampus, showing its efficacy under realistic experimental conditions. Our results thus establish NBFR-based photostimulus optimization as an important addition to an emerging computational toolkit for scalable precision optogenetics.

A fast optimization method for the propeller layout of distributed propulsion aircraft