Penalty Strategy Research Articles

Prediction of rainy-day photovoltaic power generation based on Generative Adversarial Networks and enhanced Sparrow Search Algorithm

Accurate and timely photovoltaic (PV) power forecasting is crucial for the stable operation of power systems. To address the issue of sparse PV power data on rainy days, this paper proposes the use of Wasserstein Generative Adversarial Networks with Gradient Penalty (WGAN-GP) to augment rainy-day data. A progressive gradient penalty strategy is introduced to avoid gradient vanishing. A Bidirectional Long Short-Term Memory network is employed for power forecasting, with adaptive hyperparameter optimization achieved through an improved Sparrow Search Algorithm (LSSA). The optimization capability is enhanced using a chaotic mapping strategy. Validation on data from a 1 MW PV plant demonstrates that the LSSA-optimized model achieves the best forecasting results. After augmenting the rainy-day dataset, the Mean Absolute Error and Root Mean Square Error decreased by 15.3 % and 6.15 %, respectively, indicating that WGAN-GP significantly improves forecasting accuracy on rainy days.

Integrated Optimization of Production Scheduling and Haulage Route Planning in Open-Pit Mines

In mining, deposits are divided into blocks, forming the basis for open-pit mine planning, covering production and haulage route planning. Current studies often stage optimization and lack the consideration of road capacity, leading to suboptimal solutions. A novel approach integrates production scheduling and haulage route planning through a bilevel optimization model. The upper-level model integrates ore mining constraints to establish a mixed-integer production scheduling model, minimizing haulage costs. Spatiotemporal correlation constraints for block mining are determined using a two-stage algorithm. The lower-level model incorporates road capacity, forming a haulage route optimization model based on multicommodity network flow. A solution algorithm with a distance penalty strategy facilitates feedback between the upper and lower levels, achieving optimal solutions. Tested on a real open-pit coal mine with over 5 million blocks, this approach reduces haulage costs by 10.06% compared to stage optimization. Additionally, this approach allows for adjusting haulage demand in both temporal and spatial dimensions, effectively preventing road congestion. This study advances rational mining processes and enhances the efficiency of open-pit mining haulage systems.

Multi-Label Sampling based on Label Imbalance Rate and Neighborhood Distribution

Existing multi-label classification algorithms often assume that label distribution in the training set is balanced, but practical datasets frequently exhibit significant label imbalance. This imbalance affects the learning and generalization performance of the classifiers. To address the problem of label imbalance in multi-label classification, this paper proposes a new synthetic oversampling algorithm, named Multi-Label Synthetic Oversampling based on Label Imbalance Rate and Neighborhood Distribution (MLSIN). This algorithm synthesizes new samples by considering both the imbalance rate of labels and the distribution of samples in their neighborhood, aiming to improve the classifiers performance on minority labels. The rest of this chapter first introduces the evaluation metrics for multi-label classification effectiveness. Then it defines and computes the degree of label imbalance, describes the calculation of imbalance weights, and proposes a sample type correction penalty strategy, detailing the algorithm's process for selecting base and auxiliary samples., and validates the proposed method on public datasets and summarizes the experimental results.

Consensus-based optimization for multi-objective problems: a multi-swarm approach

We propose a multi-swarm approach to approximate the Pareto front of general multi-objective optimization problems that is based on the consensus-based optimization method (CBO). The algorithm is motivated step by step beginning with a simple extension of CBO based on fixed scalarization weights. To overcome the issue of choosing the weights we propose an adaptive weight strategy in the second modeling step. The modeling process is concluded with the incorporation of a penalty strategy that avoids clusters along the Pareto front and a diffusion term that prevents collapsing swarms. Altogether the proposed K-swarm CBO algorithm is tailored for a diverse approximation of the Pareto front and, simultaneously, the efficient set of general non-convex multi-objective problems. The feasibility of the approach is justified by analytic results, including convergence proofs, and a performance comparison to the well-known non-dominated sorting genetic algorithms NSGA2 and NSGA3 as well as the recently proposed one-swarm approach for multi-objective problems involving consensus-based optimization.

Multi-patch isogeometric Kirchhoff–Love shell analysis for post-buckling of functionally graded graphene platelets reinforced composite shells

This paper develops a multi-patch isogeometric Kirchhoff–Love shell method for post-buckling of functionally graded graphene platelets reinforced composite (FG-GPLRC) cylindrical, spherical, and conoidal shell structures, which are built with single or multiple NURBS patches. A penalty strategy is employed to weakly couple nonconforming interfaces between adjacent patches. The coupling work induced by enforcing displacement continuity and rotational continuity is added to the equilibrium equation, and the corresponding stiffness matrix is derived in detail. A simplified arc-length method is utilized to capture the complex equilibrium paths including snap-through and snap-back behaviors. Five distribution patterns of the shells including uniform (UD), V-type, A-type, O-type, and X-type are considered. The cylindrical and spherical shells are subjected to concentrated loadings at central points while for conoidal shells the concentrated loadings are enforced at the center points of an edge. The post-buckling of isotropic and laminated shell structures is first studied to validate the developed formulations by comparing the obtained results with those given in existing literature. Then a series of numerical examples considering nonlinear FG-GPLRC shell problems are conducted to explore the effect of various parameters like geometric dimensions, GPL distribution patterns, and shell thickness on the mechanical performance. Finally, the post-buckling of a cylindrical shell subjected to an offset concentrated load, with extremely complicated equilibrium paths, is modeled and analyzed by using the developed multi-patch isogeometric method. The numerical results reveal that the X-type GPL distribution pattern demonstrates better performance in load–deflection responses and provides the largest buckling critical load among the five patterns. Additionally, the increase in height ratio could deteriorate the stability performance of FG-GPLRC conoidal shells.

Forest ecological long sequence missing data imputation method based on BiP-Informer

The monitoring of forest ecological data primarily relies on various types of sensors. However, long sequence of data are often missing. The commonly used long sequence missing data imputation algorithms suffer from problems such as error accumulation and lack of long-term dependencies, making it difficult to achieve accurate missing data imputation. To address this issue, this paper proposes a deep learning model dubbed BiP-Informer (Bidirectional Penalty-Informer). BiP-Informer is founded on the sparse self-attention mechanism, which adds the bidirectional information transmission mechanism to effectively capture the long-term dependencies. The bidirectional loss penalty strategy is proposed to alleviate error accumulation. The experimental results shows that the BiP-Informer outperformed commonly used methods by 25–52%, achieving the best imputation results with MAE of 0.97%, RMSE of 1.56%, and R2 of 99.25%. The BiP-Informer not only provides a new solution for sensor monitoring missing data imputation in forest areas but also has widespread application prospects.

Caching Placement Optimization Strategy Based on Comprehensive Utility in Edge Computing

With the convergence of the Internet of Things, 5G, and artificial intelligence, limited network bandwidth and bursts of incoming service requests seem to be the most important factors affecting user experience. Therefore, caching technology was introduced. In this paper, a caching placement optimization strategy based on comprehensive utility (CPOSCU) in edge computing is proposed. Firstly, the strategy involves quantifying the placement factors of data blocks, which include the popularity of data blocks, the remaining validity ratio of data blocks, and the substitution rate of servers. By analyzing the characteristics of cache objects and servers, these placement factors are modeled to determine the cache value of data blocks. Then, the optimization problem for cache placement is quantified comprehensively based on the cache value of data blocks, data block retrieval costs, data block placement costs, and replacement costs. Finally, to break out of the partial optimal solution for cache placement, a penalty strategy is introduced, and an improved tabu search algorithm is used to find the best edge server placement for cached objects. Experimental results demonstrate that the proposed caching strategy enhances the cache service rate, reduces user request latency and system overhead, and enhances the user experience.

Flow recovery from distal pressure in linearized hemodynamics: an optimal control approach

The goal of this work is to derive a reliable stable and accurate inverse problem strategy for reconstructing cardiac output blood flow entering the ascending aorta from pressure measurements at a distal site of the arterial tree, assumed here to be the descending aorta. We assume that a reduced one-dimensional model of the aorta can be linearized around its steady state, resulting in a wave system with absorbing boundary condition at the outlet. Using this model, we attempt to reconstruct the inlet flow from a pressure measurement at the distal outlet. First, we investigate the observability of the problem and prove that the inversion of the input-output operator for the flow and pressure in the space of time-periodic solutions is ill-posed of degree one. We then develop a variational approach where we minimize the discrepancy between measurements and a simulated state and penalize the error with respect to a periodic state. It is shown that the penalty strategy is convergent and provides an efficient solution for the minimization. Numerical results illustrate the robustness of our approach to noise and the potential of our method to reconstruct inlet flow from real pressure recordings during anesthesia.

Optimal cancellation penalty for competing ride-sourcing platforms under waiting time uncertainty

This paper investigates the optimal fare and penalty strategy of ride-sourcing platforms in a competing market. When ordering ride-sourcing services, passengers encounter waiting time variability due to the uncertainty in the matching and pickup processes. In some scenarios, some passengers order ride-sourcing services from multiple platforms to save waiting time, which results in subsequent order cancellations. To capture the behavior of order cancellation, this paper derives the cancellation probability based on the perceived inhomogeneous Poisson process. In view of the punishing effect and blocking effect of order cancellation penalty, this paper investigates the impact of order cancellation penalties on market equilibrium and platform revenue. We formulate a game-theoretic framework and obtain the duopoly market equilibrium to explore the optimal pricing (including penalties and fares) of different platforms. By incorporating passenger perceptions into the order cancellation behavior, this study produces insightful pricing schemes for the ride-sourcing platforms.

Multirobot Task Planning Method Based on the Energy Penalty Strategy

In multirobot task planning, the goal is to meet the multi-objective requirements of the optimal and balanced energy consumption of robots. Thus, this paper introduces the energy penalty strategy into the GA (genetic algorithm) to achieve the optimization of the task planning of multiple robots in different operation scenarios. First, the algorithm model is established, after which the objective function is constructed by taking the energy excess of the relative average energy consumption of each robot as the penalty energy, along with the total energy consumption of multiple robots. In the genetic operation, two-segment chromosome coding is used to realize the iterative optimization of the number and task sequences of robots through diversified cross and mutation operators. Then, in the task scenario with obstacles, the A* (A-Star) algorithm and GA are used to plan the optimal obstacle avoidance path and to realize the secondary optimization of the robot task sequence without changing the number of tasks. During optimization, the energy penalty strategy imposes punishment on the objective function through the size of the penalty energy, enabling the robot energy consumption to reach an equilibrium state by maintaining the total energy consumption at the minimum. Finally, the MATLAB platform is used to conduct the simulation experiments to compare with basic genetic algorithms and penalty function algorithms, after which the optimal allocation scheme and energy consumption iteration of the algorithm are analyzed under different robot numbers, task numbers, and task scenarios, and the simulation results include robot task sequences, total energy consumption, average energy consumption, and standard deviation of energy consumption.

Target-Oriented Knowledge Distillation with Language-Family-Based Grouping for Multilingual NMT

Multilingual NMT has developed rapidly, but still has performance degradation caused by language diversity and model capacity constraints. To achieve the competitive accuracy of multilingual translation despite such limitations, knowledge distillation, which improves the student network by matching the teacher network’s output, has been applied and shown enhancement by focusing on the important parts of the teacher distribution. However, existing knowledge distillation methods for multilingual NMT rarely consider the knowledge, which has an important function as the student model’s target, in the process. In this article, we propose two distillation strategies that effectively use the knowledge to improve the accuracy of multilingual NMT. First, we introduce a language-family-based approach, guiding to select appropriate knowledge for each language pair. By distilling the knowledge of multilingual teachers that each processes a group of languages classified by language families, the multilingual model overcomes accuracy degradation caused by linguistic diversity. Second, we propose target-oriented knowledge distillation, which intensively focuses on the ground-truth target of knowledge with a penalty strategy. Our method provides a sensible distillation by penalizing samples without actual targets, while additionally targeting the ground-truth targets. Experiments using TED Talk datasets demonstrate the effectiveness of our method with BLEU scores increment. Discussions of distilled knowledge and further observations of the methods also validate our results.

GA-ENs: A novel drug–target interactions prediction method by incorporating prior Knowledge Graph into dual Wasserstein Generative Adversarial Network with gradient penalty

Bipartite graph-based drug–target interactions (DTIs) prediction methods are commonly limited by the sparse structure of the graph, resulting in acquiring suboptimal node feature representations. In addition, these defective node features will also interfere with the representation quality of corresponding edge features. To alleviate the sparsity of bipartite graph and get better node representation, according to the prior Knowledge Graph (KG), we developed a novel prediction model based on Variational Graph Auto-Encoder (VGAE) combined with our proposed dual Wasserstein Generative Adversarial Network with gradient penalty strategy (dual-WGAN-GP) for generating edge information and augmenting their representations. Specifically, GA-ENs first utilized dual-WGAN-GP to fill possible edges by a prior KG containing various molecular associations knowledge when constructing a bipartite graph of known DTIs. Moreover, we utilized the KG transfer probability matrix to redefine the drug–drug and target–target similarity matrix, thus constructing the final graph adjacent matrix. Combining graph adjacent matrix with node features, we learn node representations by VGAE and augment them by utilizing dual-WGAN-GP again, thus obtaining final edge representations. Finally, a fully connected network with three layers was designed to predict potential DTIs. Extensive experiment results show that GA-ENs has excellent performance for DTIs prediction and could be a supplement tool for practical DTIs biological screening.

Construction of an epileptic seizure prediction model using a semi-supervised method of generative adversarial and long short term memory network combined with Stockwell transform

To propose a semi-supervised epileptic seizure prediction model (ST-WGAN-GP-Bi-LSTM) to enhance the prediction performance by improving time-frequency analysis of electroencephalogram (EEG) signals, enhancing the stability of the unsupervised feature learning model and improving the design of back-end classifier. Stockwell transform (ST) of the epileptic EEG signals was performed to locate the time-frequency information by adaptive adjustment of the resolution and retaining the absolute phase to obtain the time-frequency inputs. When there was no overlap between the generated data distribution and the real EEG data distribution, to avoid failure of feature learning due to a constant JS divergence, Wasserstein GAN was used as a feature learning model, and the cost function based on EM distance and gradient penalty strategy was adopted to constrain the unsupervised training process to allow the generation of a high-order feature extractor. A temporal prediction model was finally constructed based on a bi-directional long short term memory network (Bi-LSTM), and the classification performance was improved by obtaining the temporal correlation between high-order time-frequency features. The CHB-MIT scalp EEG dataset was used to validate the proposed patient-specific seizure prediction method. The AUC, sensitivity, and specificity of the proposed method reached 90.40%, 83.62%, and 86.69%, respectively. Compared with the existing semi-supervised methods, the propose method improved the original performance by 17.77%, 15.41%, and 53.66%. The performance of this method was comparable to that of a supervised prediction model based on CNN. The utilization of ST, WGAN-GP, and Bi-LSTM effectively improves the prediction performance of the semi-supervised deep learning model, which can be used for optimization of unsupervised feature extraction in epileptic seizure prediction.

Mimicry API Gateway Decision Algorithm Based on Trust Distribution

Aiming at the lack of dynamic and comprehensiveness of trust evaluation methods in current adjudication algorithms, an adjudication algorithm for mimic API gateway based on trust distribution is proposed. The algorithm comprehensively considers the direct and indirect trust relationships between equipment executives from the perspective of trust distribution, and on this basis introduces a dynamic penalty strategy, which greatly improves the dynamics and effectiveness of the algorithm. In addition, the algorithm is simulated and tested in the simulated API gateway environment, and the simulation results show that the proposed algorithm has significantly improved the reliability and security of trust evaluation compared with the previous adjudication algorithms.

Task Offloading Strategy for Ocean Based on MEC

With the development of marine business, the diversification of marine business has produced more and more delay-sensitive and computing-intensive tasks, and the demand for marine communication is becoming more and more obvious. Establish a task offloading model between ship users, select offload nodes based on the connectivity probability between ship users, in order to alleviate the problem of data congestion during transmission, use orthogonal frequency division multiple access technology to divide channels, and propose subtasks and Algorithm for subchannel matching. Ship users use solar energy to supply energy, and propose a penalty strategy for insufficient energy collection and a penalty strategy for task delay, so that the system can minimize the actual energy consumption and establish an objective function under the condition of satisfying the delay constraint. The local transmit power is optimized, and the gray wolf optimization algorithm is improved to obtain a better data offloading ratio. The simulation results of EdgeCloudSim show that, compared with other methods, the proposed method can achieve the best performance while guaranteeing the delay.

A proximal bundle method-based algorithm with penalty strategy and inexact oracles for constrained nonsmooth nonconvex optimization

In this paper, we consider a class of nonconvex nonsmooth constrained problems with inexact data. To deal with the constraints, the penalty strategy is adopted during the process to transfer the primal problem into an unconstrained problem. For inexact data, we add quadratic term with variable coefficient to each term in the unconstrained problem to keep the corresponding linearization errors nonnegative. For the modified unconstrained problem, we utilize the disaggregate technique and design corresponding cutting planes for each function. Meanwhile, the sum of the corresponding cutting planes is regarded as the cutting plane for the modified unconstrained problem and proximal bundle method is adopted to deal with the model. The preliminary numerical results show our algorithm is effective and attractive.

A large-scale consensus model to manage non-cooperative behaviors in group decision making: A perspective based on historical data

In the large-scale group decision making (LSGDM) problems, it is common that some decision makers might refuse to modify their preferences to achieve a consensus or even form an alliance to exhibit non-cooperative behaviors for their own interests. These non-cooperative behaviors might bias or hinder the consensus reaching process (CRP). In this paper, we propose a large-scale consensus model to manage non-cooperative behaviors based on the clustering method using the historical preference data of decision makers. In the proposed framework, all the historical preference data of decision makers are used for clustering to catch their features, by which the decision makers with the three defined non-cooperative behaviors will be more centralized. Then, the clusters with non-cooperative behaviors will be detected better through their preference adjustments. On this basis, the penalty strategy of the decision makers in clusters with non-cooperative behaviors would be more effective. Simulation experiments and comparison studies are presented to demonstrate the validity of the proposed consensus framework against traditional frameworks for coping with non-cooperative behaviors.

Dynamic On-Demand Crowdshipping Using Constrained and Heuristics-Embedded Double Dueling Deep Q-Network

This paper proposes a deep reinforcement learning (DRL)-based approach to the dynamic on-demand crowdshipping problem in which requests constantly arrive in a crowdshipping system for pickup and delivery within limited time windows. The request pickup and delivery are performed by crowdsourcees, who are ordinary people dynamically arriving in and leaving the crowdshipping system, and dedicating their limited and heterogeneous available time and carrying capacity to crowdshipping. In return, crowdsourcees get paid by the delivery service provider who periodically assigns requests to crowdsourcees in the course of a day to minimize shipping cost. We adopt heuristics-embedded Deep Q-Network (DQN) algorithms that incorporate double and dueling structures, to train DRL agents. The idea of heuristics-embedded training is conceived by designing an elaborate action space where several refined local search heuristics are embedded to direct the specific action to take once an action type is chosen by DRL, with the purpose of preserving tractability of DRL training. To tackle the hard constraints pertaining to crowdsourcee and request time windows, we propose and integrate three new strategies (feasibility enforced local search, multiple schedules with different penalties, and exponential penalty) as part of the DRL training and testing. Extensive numerical analysis is conducted and shows that Double Dueling DQN with the exponential penalty strategy demonstrates the best performance. We compare the performance of the agent trained by Double Dueling DQN with conventional heuristic approaches, and find that the agent yields total shipping costs that are on average 24–37% lower than the conventional heuristic approaches. For problem instances that can be solved to optimality, the optimality gap using the trained agent is also quite small, in the range of 3–7%. Moreover, the trained agent is robust to stationary/non-stationary demand patterns. Lastly, our approach is further compared with a recent study that uses heuristics-embedded DQN, and shows superior performance (total shipping costs on average 19% lower) as a result of several differences.

Incentives to Enhance Production Reliability against Disruption: Cost-Sharing vs. Penalty

Two kinds of incentive strategies, cost-sharing and penalty, are examined in dealing with production disruption, with consideration of production process reliability as an endogenous factor for a two-echelon supply chain. Based on the Stackelberg game framework, we derive the optimal decisions of supply chain partners and compare their expected profits with different strategies. Considering the uncertain demand and the retailer’s preference against the risk, we further analyze how the partners’ decisions and the retailer’s expected profit are influenced by the feature of loss aversion. From theoretical analysis and numerical experiments, we find that: (1) overall, a penalty strategy dominates that of cost-sharing for the retailer, whereas the reverse applies with respect to the manufacturer; (2) a penalty strategy may outperform a cost-sharing strategy for the whole supply chain, depending on demand; and (3) a reasonable aversion against risk can help the retailer to achieve a more robust result when a penalty strategy is adopted under volatile and unpredictable demand.

PregGAN: A prognosis prediction model for breast cancer based on conditional generative adversarial networks

Background and Objective: Generative adversarial network (GAN) is able to learn from a set of training data and generate new data with the same characteristics as the training data. Based on the characteristics of GAN, this paper developed its capability as a tool of disease prognosis prediction, and proposed a prognostic model PregGAN based on conditional generative adversarial network (CGAN).Methods: The idea of PregGAN is to generate the prognosis prediction results based on the clinical data of patients. PregGAN added the clinical data as conditions to the training process. Conditions were used as the input to the generator along with noises. The generator synthesized new samples using the noises vectors and the conditions. In order to solve the mode collapse problem during PregGAN training, Wasserstein distance and gradient penalty strategy were used to make the training process more stable.Results: In the prognosis prediction experiments using the METABRIC breast cancer dataset, PregGAN achieved good results, with the average accurate (ACC) of 90.6% and the average AUC (area under curve) of 0.946.Conclusions: Experimental results show that PregGAN is a reliable prognosis predictive model for breast cancer. Due to the strong ability of probability distribution learning, PregGAN can also be used for the prognosis prediction of other diseases.