A CFD-driven supervised learning framework for rapid DLD microfluidic chip optimization in CTC separation.

Purpose As cities continue to grow rapidly, urban open spaces are increasingly recognized for their contribution to sustainability, well-being and environmental quality. Yet, they remain underexplored compared to built environments, particularly regarding spatial metrics such as sky view factor (SVF) and visibility, which influence perceived thermal comfort, safety and openness. These metrics are fully morphology-based and ideal for early-stage planning, but current tools often lack support for localized, interpretable optimization. This study focuses on green spaces such as parks and introduces an AI-driven framework for optimizing SVF and visibility, to enhance the usability and spatial performance of these spaces. Design/methodology/approach The framework combines Machine Learning Models (MLMs), explainable AI techniques such as SHapley Adaptive Explanations (SHAP), and Counterfactual Explanations (CFXs) to enable localized, interpretable, and efficient optimization. A dataset of 1,152 simulated urban block configurations based on Tehran's morphological patterns was used to train and test five predictive models. SHAP was used to assess the contribution of each feature, while CFXs proposed minimal design interventions to optimize spatial performance. Findings The results show that park area and building height on the east and west sides are critical factors for SVF, while southern building distance and building width most strongly affect visibility. XGBoost achieved an R2 of 0.91 for SVF prediction and F1-scores up to 0.89 for visibility classification. The CFX approach produced optimized results in approximately 1 min with an average RMSE of 5%, whereas genetic algorithms required 15–30 min to converge. This represents a ∼90% reduction in computation time. Originality/value Unlike global optimization methods, which are computationally intensive and impractical for localized adjustments, this framework supports incremental design improvements with lower computational costs and greater flexibility. This study offers a novel application of CFXs for spatial metric optimization in urban open space design, bridging a gap in current literature. The integration of explainable AI not only improves transparency but also supports real-time decision-making in sustainable urban design.

<p><b>The client-centric multi-cloud has become a popular cloud ecosystem because it allows enterprise users to share the workload across multiple cloud providers to achieve high-quality services with lower operation costs and higher application resilience. From the perspective of application providers, the location of cloud resources for application deployment significantly impacts the deployment cost and performance of applications, e.g., request response time. This gives rise to the problem of location-aware application deployment in multi-cloud to select suitable cloud resources from widely distributed multi-cloud data centers to balance the cost and performance. Existing research works did not pay full attention to the key impact of the location for application deployment. Therefore, it is urgent to study the problem both theoretically and in practice. In this thesis, innovative optimization methods and machine learning techniques are proposed for three common scenarios, namely composite application deployment, application replication and deployment, and elastic application deployment.</b></p> <p>First, this thesis studies the composite application deployment problem with the goal to minimize the average response time of composite applications subject to a budget constraint. We propose a Hybrid Genetic Algorithm (GA)-based approach, i.e., H-GA, for solving the problem with an extremely large search space. H-GA features a newly designed and domain-tailored service clustering algorithm, repair algorithm, solution representation, population initialization, and genetic operators. Experiments show that H-GA can outperform significantly several state-of-the-art approaches, achieving up to about 8% performance improvement in terms of response time, and 100% budget satisfaction in the meantime.</p> <p>Second, this thesis studies the application replication and deployment problem with the goal to minimize the total deployment cost of all application replicas subject to a stringent requirement on average response time. We propose two approaches under different optimization frameworks to solve the problem. With user requests dispatched to the closest application replicas, we develop an approach under a GA framework for Application Replication and Deployment (ARD), i.e., GA-ARD. GAARD features problem-specific solution representation, fitness measurement, and population initialization, which are effective to optimize the deployment of application replicas in multi-cloud. The experiments show that GA-ARD outperforms common application replication and placement strategies in the industry. With user requests flexibly dispatched among different application replicas, we develop another approach under a two-stage optimization framework, i.e., MCApp. MCApp can optimize both replica deployment and request dispatching by combining mixed-integer linear programming with domain-tailored large neighborhood search. Our experiments show that MCApp can achieve up to 25% reduction in total deployment cost compared with several recently developed approaches.</p> <p>Third, this thesis studies the elastic application deployment problem to minimize the deployment cost over a time span such as a billing day while satisfying the constraint on average response time. The goal of adapting resources for application deployment in response to dynamic and distributed workloads motivates us to adopt deep reinforcement learning (DRL) techniques. The proposed approach, namely DeepScale, applies a deep Q-network (DQN) to capture the optimal scaling policy that can perform online resource scaling. DeepScale also includes a long short-term memory-based prediction model to allow the DQN to consider predicted future requests while making cost-effective scaling decisions. Besides, we design a penalty-based reward function and a safety-aware action executor to ensure that any scaling decisions made by DRL can satisfy the response time constraint. The experiments show that DeepScale can significantly reduce the deployment cost of applications compared with the state-of-the-art baselines, including Amazon auto-scaling service and recently proposed RL-based algorithms. In the meanwhile, DeepScale can effectively satisfy the constraint on the average response time.</p> <p>In summary, this thesis studies three new problems for location-aware application deployment in multi-cloud. We propose four novel approaches under different optimization and machine learning frameworks, i.e., HGA, GA-ARD, MCApp, and DeepScale, for solving these problems. New constraint handling techniques are developed to satisfy the practical deployment requirements of enterprise applications.</p>

The surface figure precision, weight, and dynamic performance of spaceborne primary mirrors depend on mirror structure parameters, which are usually optimized to improve the overall performance. To realize rapid multi-objective design optimization of a primary mirror with multiple apertures, a fully parameterized primary mirror structure is established. A surrogate model based on a hybrid of improved particle swarm optimization (IPSO), adaptive genetic algorithm (IAGA), and optimized back propagation neural network (IPSO-IAGA-BPNN) is developed to replace optomechanical simulation with its high computational cost. In this model, a self-adaptive inertia weight and a modified genetic operator are introduced into the particle swarm optimization (PSO) and adaptive genetic algorithm (AGA), respectively. The connection parameters of BPNN are optimized by the IPSO-IAGA algorithm for global searching capability. Further, the proposed IPSO-IAGA-BPNN, based on a rapid multi-objective optimization framework for a fully parameterized primary mirror structure, is established. Moreover, in addition to the proposed IPSO-IAGA-BPNN model, the Kriging, RSM, BPNN, GA-BPNN, PSO-BPNN, and PSO-GA-BPNN models are also analyzed as contrast models. The comparison results indicate that the predicted value obtained by IPSO-IAGA-BPNN is superior to the six other surrogate models since its mean absolute percentage error is less than 3% and its R2 is more than 0.99. Finally, we present a Pareto-optimal primary mirror design and implement it through three optimization methods. The verification results show that the proposed method predicts mirror structural performance more accurately than existing surrogate-based methods, and promotes significantly superior computational efficiency compared to the conventional integration-based method.

In the paper, a framework for computationally-efficient design optimization of compact rat-race couplers (RRCs) is discussed. A class of hybrid RRCs with variable operating conditions is investigated, whose size reduction is obtained by replacing ordinary transmission lines with compact microstrip resonant cells (CMRCs). Our approach employs a bottom-up design strategy leading to the development of compact RRCs through rapid design optimization of its building blocks and a subsequent fine tuning to account for parasitic cross-coupling effects. The fundamental component of the proposed method is an inverse CMRC surrogate model, covering a wide range of cell electrical parameters, and enabling a convenient adjustment of coupler bandwidth. Having the surrogate model established, it is possible to produce close-to-optimum CMRC dimensions at a negligible computational cost. The subsequent correction step requires only up to two electromagnetic simulations of the CMRC. The proposed method is demonstrated by designing an RRC for several operational bandwidths. Experimental results are also provided.

A machine learning framework for real-time inverse modeling and multi-objective process optimization of composites for active manufacturing control

The variable-camber continuous trailing-edge flap is a novel control surface technology designed to serve as an aeroelastic wing shaping control device. Capable of minimizing drag throughout an aircraft flight envelope, use of the variable-camber continuous trailing-edge flap technology is examined on a flexible-wing transport aircraft configuration known as the NASA generic transport model. Drag reduction studies at various flight conditions are conducted in order to examine the performance of different flap configurations with selected camber deflection profiles. Rapid optimization of the design space is achieved by using aerodynamic–structural modeling based on a vortex lattice method coupled with transonic small disturbance and integral boundary-layer solutions. The begin cruise optimization results show that a parabolic flap deflection profile configuration with three-cambered-segments achieves the largest drag reduction of 8.4% as compared to a clean wing configuration. The configuration is used to run additional optimizations measuring performance benefits at different flight conditions including 20% fuel end cruise, a design lift coefficient, and a Mach overspeed condition. The optimization framework and initial results produce insights that can be used to refine the optimization of variable-camber continuous trailing-edge flap configurations with higher-fidelity aerodynamic tools.

The smart parking system is one of the most important problems in smart cities. Recently, an increasing number of sensors installed in parking spaces have provided big spatio-temporal data that be used to analyze parking situations in the city and help parking officers monitor parking violations. The traveling officer problem was customized to formulate a path-finding problem that aims to maximize the probability of catching overstayed cars before they leave. One of the challenges is to extract effective features from the big spatio-temporal data and provide a data-driven solution to replace conventional solutions such as a simple rule-based system or single optimization methods. In this article, we propose a seamless end-to-end learning and optimization framework that combines the long short-term memory auto-encoder neural network, clustering, and path-finding methods to solve the traveling officer problem. Our extensive comparison experiments on a large-scale real-world dataset have shown that our proposed solution outperforms any other single-step or optimization methods.

Coordinately embedding machine learning and optimization in the Monte Carlo simulation is proposed as a new framework for reliability evaluation, which is becoming more challenging due to the joining of variability and intermittency of the renewable generation and components’ ageing. As a machine learning method, the dynamic Bayesian belief network predicts the renewable outputs by generating their probability distributions through historical data to overcome the defect of rarely grasping the low-probability events in the traditional methods, which either predict a single-point value or presume a parametric probability distribution. As the internal operation module of the framework, the rolling-horizon unit commitment maintains the complexity of the operation model and meantime punctually updates the predicted renewable generation and the components’ ageing. The dynamic Bayesian belief network and the rolling-horizon unit commitment traverse time step after time step throughout the horizon, and thus form one bid of the Monte Carlo simulation. In each time step, the dynamic Bayesian belief network extracts the probability distributions of the renewable generation, from which the boundary of the commitment is sampled, and the impact of scheduling the generators on their ageing is then accumulated from the commitment. The proposed framework has its effectiveness proved on a microgrid.

The exponential growth of data-intensive applications and heterogeneous network architectures has increased the demand for intelligent systems capable of predicting network performance and optimizing data flow in real time. Traditional static models are inadequate for addressing dynamic network conditions, latency variations, and bandwidth fluctuations. This paper presents a comprehensive review of existing methodologies and proposes a conceptual machine learning framework for predictive network performance and data flow optimization. The framework integrates supervised and reinforcement learning models with deep neural architectures to forecast traffic patterns, congestion probabilities, and throughput variations. Furthermore, the study explores hybrid approaches that combine network telemetry data, software-defined networking (SDN), and edge intelligence for adaptive traffic routing and self-healing network behaviors. Comparative analyses of feature selection techniques, model training strategies, and optimization algorithms are also discussed to highlight trade-offs in scalability, accuracy, and computational efficiency. The paper concludes by identifying emerging trends such as federated learning, graph neural networks, and explainable AI in predictive network management. By synthesizing insights from current literature and proposing a unified framework, this study contributes to the advancement of intelligent network operations, enabling proactive maintenance, reduced latency, and improved resource utilization across next-generation communication systems.

Chip-integrated, two-dimensional high performance liquid chromatography is introduced to monitor enantioselective continuous micro-flow synthesis. The herein described development of the first two-dimensional HPLC-chip was realized by the integration of two different columns packed with reversed-phase and chiral stationary phase material on a microfluidic glass chip, coupled to mass spectrometry. Directed steering of the micro-flows at the joining transfer cross enabled a heart-cut operation mode to transfer the chiral compound of interest from the first to the second chromatographic dimension. This allows for an interference-free determination of the enantiomeric excess by seamless hyphenation to electrospray mass spectrometry. The application for rapid reaction optimization at micro-flow conditions is exemplarily shown for the asymmetric organocatalytic continuous micro-flow synthesis of warfarin.

Engineering electron correlations in quantum dot arrays demand navigation of high-dimensional, non-convex parameter spaces, where hole doping fundamentally alters the physics. We present a rigorous comparative study of two control paradigms for the 1-hole of half-filled Hubbard model: (i) systematic physics-guided design and (ii) autonomous deep reinforcement learning (RL) with geometry-aware neural architectures. While systematic analysis reveals key design principles—such as field-induced localization for trapping the mobile hole—it is computationally intractable for optimization. We demonstrate that an autonomous RL agent, benchmarked across five 3D lattices (tetrahedron to FCC), achieves human-competitive accuracy (R2 > 0.97) and 95.5% success on held-out tasks. Critically, the RL agent achieves this performance with 103−4× greater sample efficiency than grid search and outperforms other black-box optimization methods. Transfer learning demonstrates 91% few-shot generalization to unseen geometries. This work establishes autonomous RL as a viable, highly efficient framework for rapid optimization and non-obvious strategy discovery in complex quantum systems.

ROL-PEBBL is a C++, MPI-based parallel code for mixed-integer PDE-constrained optimization (MIPDECO). In these problems we wish to optimize (control, design, etc.) physical systems, which must obey the laws of physics, when some of the decision variables must take integer values. ROL-PEBBL combines a code to efficiently search over integer choices (PEBBL = Parallel Enumeration Branch-and-Bound Library) and a code for efficient nonlinear optimization, including PDE-constrained optimization (ROL = Rapid Optimization Library). In this report, we summarize the design of ROL-PEBBL and initial applications/results. For an artificial source-inversion problem, finding sources of pollution on a grid from sparse samples, ROL-PEBBLs solution for the nest grid gave the best optimization guarantee for any general solver that gives both a solution and a quality guarantee.

In this study, an efficient optimization framework was developed to determine the parameters of through‐silicon vias and the layout of heat sources in three‐dimensional integrated circuits (3D ICs), employing an artificial neural network (ANN) reduced‐order model in conjunction with a Bayesian optimization (BO) algorithm. The proposed method effectively predicts the temperature distribution in 3D ICs and refines their thermal parameters, offering solutions to thermal management challenges. Latin hypercube sampling was utilized for data sampling, enhancing the previously established rapid thermal analysis method through parameterization of heat source locations. The temperature distribution data for varying hotspot locations in 3D ICs were fitted using an appropriately defined objective function, leading to the development of a reduced‐order ANN model that accelerates temperature prediction. The computational results demonstrate that the neural network model exhibits a deviation in predicted values of less than 2%, and the coefficient of determination R2 approximately 0.93, underscoring high predictive accuracy. Additionally, the optimization outcomes and the efficiency of the selected BO algorithm were thoroughly evaluated. Notably, the BO algorithm achieved the global optimum in just 4.07 s across 250 iterations, demonstrating an effective power distribution strategy for the 3D ICs model.

The hot stamping process of aluminum alloys involves multiple parameters, including blank holder force, stamping speed, die temperature, and friction coefficient. Traditional methods often fail to capture the nonlinear interactions among these parameters. This study proposes an optimization framework that integrates BP neural networks with genetic algorithms (GA), while six bio-inspired algorithms—Grey Wolf Optimization (GWO), Sparrow Search Algorithm (SSA), Crested Porcupine Optimizer (CPO), Grey lag Goose Optimization (GOOSE), Dung Beetle Optimizer (DBO), and Parrot Optimizer (PO)—were employed to optimize the network hyperparameters. Comparative results show that all optimized models outperformed the baseline BP model (R2 = 0.702, RMSE = 0.106, MAPE = 20.8%). The PO-BP achieved the best performance, raising R2 by 27.3% and reducing MAPE by 27.1%. Furthermore, combining GA with the PO-BP model yielded optimized process parameters, reducing the maximum thinning rate to 17.0% with only a 1.16% error compared with experiments. Overall, the proposed framework significantly improves prediction accuracy and forming quality, offering an efficient solution for rapid process optimization in intelligent manufacturing of aluminum alloy automotive parts.

Background/Objectives: The manual tuning of exoskeleton control parameters is tedious and often ineffective for adapting to individual users. Human-in-the-loop (HIL) optimization offers an automated approach, but existing methods typically rely on metabolic cost, which requires prolonged data collection times of at least 60 s. Surface electromyography (EMG) signals, as an alternative, enable faster optimization with reduced data acquisition times. Methods: This study develops a rapid EMG-based HIL Bayesian optimization framework to tune hip exoskeleton controllers for assisting free leg swinging. Eight participants are asked to perform leg swinging at two frequencies with assistance from a hip exoskeleton. EMG signals from four sensors, representing muscle activity during forward and backward swings, are dynamically processed into cost functions. Bayesian optimization with Gaussian processes tunes four controller parameters using an expected improvement acquisition function. Optimization outcomes are validated against no device, zero torque, and general control baselines. Results: Optimization converges within an average of 142 s with a standard deviation of 24 s across all participants. The controller yields muscle activity reductions of 16.1% (p < 0.001) compared to no device, 21.7% (p < 0.001) versus zero torque, and 15.1% (p < 0.001) versus general control. EMG-based tuning is faster than metabolic-cost-based methods and perceived as less effortful, with Borg scale reductions of up to 39.5%. Conclusions: EMG-based HIL optimization significantly enhances controller tuning speed and effectiveness, demonstrating its potential for scalable and user-specific exoskeleton applications.