Architecture-based Model Research Articles

On Estimating the Feasible Solution Space of Multi-objective Testing Resource Allocation

The multi-objective testing resource allocation problem (MOTRAP) is concerned on how to reasonably plan the testing time of software testers to save the cost and improve the reliability as much as possible. The feasible solution space of a MOTRAP is determined by its variables (i.e., the time invested in each component) and constraints (e.g., the pre-specified reliability, cost, or time). Although a variety of state-of-the-art constrained multi-objective optimisers can be used to find individual solutions in this space, their search remains inefficient and expensive due to the fact that this space is very tiny compared to the large search space. The decision maker may often suffer a prolonged but unsuccessful search that fails to return a feasible solution. In this work, we first formulate a heavily constrained MOTRAP on the basis of an architecture-based model, in which reliability, cost, and time are optimised under the pre-specified multiple constraints on reliability, cost, and time. Then, to estimate the feasible solution space of this specific MOTRAP, we develop theoretical and algorithmic approaches to deduce new tighter lower and upper bounds on variables from constraints. Importantly, our approach can help the decision maker identify whether their constraint settings are practicable, and meanwhile, the derived bounds can just enclose the tiny feasible solution space and help off-the-shelf constrained multi-objective optimisers make the search within the feasible solution space as much as possible. Additionally, to further make good use of these bounds, we propose a generalised bound constraint handling method that can be readily employed by constrained multi-objective optimisers to pull infeasible solutions back into the estimated space with theoretical guarantee. Finally, we evaluate our approach on application and empirical cases. Experimental results reveal that our approach significantly enhances the efficiency, effectiveness, and robustness of off-the-shelf constrained multi-objective optimisers and state-of-the-art bound constraint handling methods at finding high-quality solutions for the decision maker. These improvements may help the decision maker take the stress out of setting constraints and selecting constrained multi-objective optimisers and facilitate the testing planning more efficiently and effectively.

Deep Learning based Semantic Segmentation for Buildings Detection from Remote Sensing Images

Building extraction from remote sensing images is the process of automatically identifying and extracting the boundaries of buildings from high-resolution aerial or satellite images. The extracted building footprints can be used for a variety of applications, such as urban planning, disaster management, city development, land management, environmental monitoring, and 3D modeling. The results of building extraction from remote sensing images depend on several factors, such as the quality and resolution of the image and the choice of algorithm.The process of building extraction from remote sensing images typically involves a series of steps, including image pre-processing, feature extraction, and classification. Building extraction from remote sensing images can be challenging due to factors such as varying building sizes and shapes, shadows, and occlusions. However, recent advances in deep learning and computer vision techniques have led to significant improvements in the accuracy and efficiency of building extraction methods. This research presents a deep learning semantic segmentation architecture-based model for developing building detection from high resolution remote sensing images. The open-source Massachusetts dataset is used to train the suggested UNet architecture. The model is optimized using the RMSProp algorithm with a learning rate of 0.0001 for 100 epochs. After 1.52 hours of training on Google Colab the model achieved an 83.55% F1 score, which indicates strong precision and recall.

Manifold attention-enhanced multi-domain convolutional network for decoding motor imagery intention

Research in brain-computer interface (BCI), particularly in brain intention decoding, has made significant strides relying on the remarkable capabilities of deep learning (DL). However, many current DL framework-based models struggle to effectively utilize complementary information present in various domains of electroencephalographic (EEG) signals. This limitation may lead to suboptimal performance, particularly exacerbated by the presence of noise and artifacts. Therefore, this study proposed a manifold attention-based multi-domain convolutional neural network, known as MAMCNet, to benefit from complementary strengths across different domains of EEG signals. Specifically, the EEG signals were divided into multi-band data representations based on frequency. Multiple spatiotemporal convolutional blocks were then devised to simultaneously extract the spatiotemporal feature maps from these sub-band components. To enhance the model's robustness, these maps were mapped onto the Riemannian manifold. Additionally, manifold attention was utilized to emphasize salient spatiotemporal feature maps with attention focused on the target. A frequency convolutional block was specifically designed to merge optimized feature maps from different bands, extracting temporospatiofrequency features in EEG signals comprehensively. The experimental results on two datasets demonstrated that the proposed model outperformed the mainstream DL architecture-based model. These results suggest that the proposed model could offer an alternative scheme for facilitating BCI applications.

Deep Learning for Sarcasm Identification in News Headlines

Sarcasm is a mode of expression whereby individuals communicate their positive or negative sentiments through words contrary to their intent. This communication style is prevalent in news headlines and social media platforms, making it increasingly challenging for individuals to detect sarcasm accurately. To mitigate this challenge, developing an intelligent system that can detect sarcasm in headlines and news is imperative. This research paper proposes a deep learning architecture-based model for sarcasm identification in news headlines. The proposed model has three main objectives: (1) to comprehend the original meaning of the text or headlines, (2) to learn the nature of sarcasm, and (3) to detect sarcasm in the text or headlines. Previous studies on sarcasm detection have utilized datasets of tweets and employed hashtags to differentiate between ordinary and sarcastic tweets depending on the limited dataset. However, these datasets were prone to noise regarding language and tags. In contrast, using multiple datasets in this study provides a comprehensive understanding of sarcasm detection in online communication. By incorporating different types of sarcasm from the Sarcasm Corpus V2 from Baskin Engineering and sarcastic news headlines from The Onion and HuffPost, the study aims to develop a model that can generalize well across different contexts. The proposed model uses LSTM to capture temporal dependencies, while the proposed model utilizes a GlobalMaxPool1D layer for better feature extraction. The model was evaluated on training and test data with an accuracy score of 0.999 and 0.925, respectively.

New Reliability-Driven Bounds for Architecture-Based Multi-Objective Testing Resource Allocation

The multi-objective testing resource allocation problem (MOTRAP) aims at seeking a good trade-off between system reliability, testing cost, and testing time, which is of significant importance to facilitate the testing planning. Yet most studies focus on the time constraint but rarely consider the practical reliability requirement. In this work, we address MOTRAP on an architecture-based model (ABM) with the personalized preference over reliability. More specifically, we first present a reliability-constrained MOTRAP model on the basis of ABM and illustrate how to use this model for real-world systems. Then, to leverage the problem's knowledge, we develop new lower and upper bounds on testing time invested in different components from both theoretical and algorithmic perspectives on the basis of the Lagrange multiplier and half-interval search. Importantly, these new derived bounds have strong implications due to the fact that they can be easily employed by optimizers as the limits of variables to prune the search space to the region of interests of the decision maker and locate feasible solutions with the expected reliability. Finally, we evaluate the proposed bounds in popular multi-objective optimizers for MOTRAP on application and empirical cases. Experimental results demonstrate that our new bounds practically improve the search performance of optimizers, and decision makers can easily combine these new bounds with off-the-shelf optimizers to find higher-quality solutions that they are interested in, which greatly soothes away stress on optimizer and solution selections of decision makers.

Deep transfer learning model for disease identification in wheat crop

Wheat rusts, caused by pathogenic fungi, are responsible for significant losses in Wheat production. Leaf rust can cause around 45–50% crop loss, whereas stem and stripe rust can cause up to 100% crop loss under suitable weather conditions. Early treatment is crucial in reducing yield loss and improving the effectiveness of phytosanitary measures. In this study, an EfficientNet architecture-based model for Wheat disease identification is proposed for automatically detecting major Wheat rusts. We prepared a dataset, referred to as WheatRust21, consisting of 6556 images of healthy and diseased leaves from natural field conditions. We attempted several classical CNN-based models such as VGG19, ResNet152, DenseNet169, InceptionNetV3, and MobileNetV2 for Wheat rust disease identification and obtained accuracy ranging from 91.2 to 97.8%. To further improve accuracy, we experimented with eight variants of EfficientNet architecture and discovered that our fine-tuned EfficientNet B4 model achieved a testing accuracy of 99.35%, a result that has not been reported in the literature so far to the best of our knowledge. This model can be easily integrated into mobile applications for use by stakeholders for image-based wheat disease identification in field conditions.

An explainable COVID-19 detection system based on human sounds.

Acoustic signals generated by the human body have often been used as biomarkers to diagnose and monitor diseases. As the pathogenesis of COVID-19 indicates impairments in the respiratory system, digital acoustic biomarkers of COVID-19 are under investigation. In this paper, we explore an accurate and explainable COVID-19 diagnosis approach based on human speech, cough, and breath data using the power of machine learning. We first analyze our design space considerations from the data aspect and model aspect. Then, we perform data augmentation, Mel-spectrogram transformation, and develop a deep residual architecture-based model for prediction. Experimental results show that our system outperforms the baseline, with the ROC-AUC result increased by 5.47%. Finally, we perform an interpretation analysis based on the visualization of the activation map to further validate the model.

Siamese-Based Architecture for Cross-Lingual Plagiarism Detection in English-Hindi Language Pairs.

The cross-lingual plagiarism detection (CLPD) is a challenging problem in natural language processing. Cross-lingual plagiarism is when a text is translated from any other language and used as it is without proper acknowledgment. Most of the existing methods provide good results for monolingual plagiarism detection, whereas the performances of existing methods for the CLPD are very limited. The reason for this is that it is difficult to represent the text from two different languages in a common semantic space. In this article, a novel Siamese architecture-based model is proposed to detect the cross-lingual plagiarism in English-Hindi language pairs. The proposed model combines the convolutional neural network (CNN) and bidirectional long short-term memory (Bi-LSTM) network to learn the semantic similarity among the cross-lingual sentences for the English-Hindi language pairs. In the proposed model, the CNN model learns the local context of words, whereas the Bi-LSTM model learns the global context of sentences in forward and backward directions. The performances of the proposed models are evaluated on the benchmark data set, that is, Microsoft paraphrase corpus, which is converted in the English-Hindi language pairs. The proposed model outperforms other models giving 67%, 72%, and 67% weighted average precision, recall, and F1-measure scores. The experimental results show the effectiveness of the proposed models over the baseline models because the proposed model is very efficient in representing the cross-lingual text very efficiently.

Efficient anomaly recognition using surveillance videos.

Smart surveillance is a difficult task that is gaining popularity due to its direct link to human safety. Today, many indoor and outdoor surveillance systems are in use at public places and smart cities. Because these systems are expensive to deploy, these are out of reach for the vast majority of the public and private sectors. Due to the lack of a precise definition of an anomaly, automated surveillance is a challenging task, especially when large amounts of data, such as 24/7 CCTV footage, must be processed. When implementing such systems in real-time environments, the high computational resource requirements for automated surveillance becomes a major bottleneck. Another challenge is to recognize anomalies accurately as achieving high accuracy while reducing computational cost is more challenging. To address these challenge, this research is based on the developing a system that is both efficient and cost effective. Although 3D convolutional neural networks have proven to be accurate, they are prohibitively expensive for practical use, particularly in real-time surveillance. In this article, we present two contributions: a resource-efficient framework for anomaly recognition problems and two-class and multi-class anomaly recognition on spatially augmented surveillance videos. This research aims to address the problem of computation overhead while maintaining recognition accuracy. The proposed Temporal based Anomaly Recognizer (TAR) framework combines a partial shift strategy with a 2D convolutional architecture-based model, namely MobileNetV2. Extensive experiments were carried out to evaluate the model’s performance on the UCF Crime dataset, with MobileNetV2 as the baseline architecture; it achieved an accuracy of 88% which is 2.47% increased performance than available state-of-the-art. The proposed framework achieves 52.7% accuracy for multiclass anomaly recognition on the UCF Crime2Local dataset. The proposed model has been tested in real-time camera stream settings and can handle six streams simultaneously without the need for additional resources.

Architecture dependent strengthening mechanisms in graphene/Al heterogeneous lamellar composites

An architecture-based model revealing the architecture-strengthening mechanism correlations of Graphene/aluminum (GR/Al) composites with heterogeneous lamellar (HL) architecture is proposed. The microstructural features such as matrix grain size, shape, and orientation in HL GR/Al composites are drivers for the dislocation storage and strain hardening ability. The proposed model allows to describe the role of the HL design factors affecting the microstructural architecture in GR/Al composite and the strengthening mechanisms of the material. The improved mechanical properties are linked to strain hardening ability, the high dislocation storage capability of the coarse grain bands (CGBs) as well as the generation of further geometrically necessary dislocations (GNDs) in the GR/Al interfaces. Moreover, the optimal-alignment of the structural elements and GRs not only causes a combined effect of isotropic and kinematic strain hardening but also leads to higher levels of back stress and back stress hardening, respectively. Actually, the significant effect of the interfacial dislocations including GNDs on the internal stresses especially back stress, which leads to the generation of extra strain hardening is underlined. In a word, the proposed model, which is based on the dislocation induced strengthening mechanisms, enables to guide the processing-microstructure design.

Terrain surveillance system with drone and applied machine vision

Road accidents are a major contribution to the Annual death rates all over the world. India, ranks first globally in the number of fatalities from road accidents. According to the Ministry of Roads & Transportation, India saw over 440,000 road accidents in 2019. As a result, over 150,000 lives were lost. Poor road conditions contribute to these directly and indirectly. In India, safety standards and conditions of roads are maintained by local bodies in a given area of jurisdiction. While there have been several attempts at improving the quality of roads, weren’t instrumental in giving proper results [42]. A recent study suggested that Artificial Intelligence (AI) might help achieve the goals. Some of the AI applications have had better results when powered with Computer Vision. While computer vision has been previously used to identify faults in roads, it is not widely implemented or made available for public use. Road inspection still largely remains a time-consuming manual task, hindering the maintenance process in most cities. Moreover, being unaware of unattended faults on roads is often the cause of road accidents, especially in rough weather conditions that make it impossible for drivers to visually gauge any dangers on their route. The proposed model uses a transfer-learning approach; using Mask R-CNN in identifying the defects at an instance level segmentation. As adding this, it requires less labelling and an additional mask helps in blocking out extra noise around the images. This paper trains a Mask R-CNN architecture-based model to identify potholes, discontinuous roads, blind spots, speed bumps, and the type of road--gravel, concrete, asphalt, tar, or mud--with a dataset of images obtained from a drone. The model is further trained to create depth maps and friction estimates of the roads being surveyed. Once trained, the model is tested on a drone-captured live feed of roads in Chennai, India. The results, once sufficiently accurate, will be implemented in a practical application to help users assess road conditions on their path.

Postbuckling response of unitized stiffened textile composite panels: Computational modeling

A computational investigation of unitized stiffened textile composite panels that are manufactured from triaxially braided textile composites and loaded in axial compression into the postbuckling regime is presented. The stiffened panels are unitized because the textile nature of the constituent material, in conjunction with an alternative stiffener design, form a near-net shape one-piece stiffened panel structure where the textile laminae that form the skin section of the panel are also used, without discontinuity of material, to form the stiffener. A multiscale computational model is presented where the architecture-dependent behavior of a discretely modeled mesoscale representative volume element is passed up to the orthotropic homogenized macroscale stiffened panel shell model. The mesoscale model is automatically executed based on input from the macroscale shell model. Progressive damage initiation and propagation is accomplished by implementing a crack band model in the macroscale model to simulate a material softening behavior. The experimental results are compared to the computational multiscale framework analysis results where a mesoscale triaxially braided textile composite architecture-based model serves as the damage initiation criterion for a smeared progressive damage crack band model in the orthotropic homogenized stiffened panel macroscale shell model.

Postbuckling response of unitized stiffened textile composite panels: Experiments

An experimental investigation of the compressive response of unitized stiffened textile composite panels that are manufactured using two-dimensional triaxially braided composites loaded in axial compression well into the postbuckling regime is presented. The stiffened panels are unitized because the constituent material was a textile laminate with an alternative method to form the stiffener geometry to a near net shape one-piece stiffened panel structure, thus removing the mechanism of stiffener debonding (separation) that is seen in traditional stiffened panels. Textile laminae that form the skin section of each panel were used, without discontinuity of material, to form the stiffener. Eight stiffened panels were manufactured using two different triaxially braided composite textiles. All eight manufactured stiffened panels withstood significant loading in the postbuckling regime prior to collapse, and nonlinear geometric and material effects were observed. Stiffener separation was not observed. Textile architecture-dependent matrix cracking was observed with local matrix cracks strongly aligned to the constituent triaxially braided composite textile. Tow buckling was observed as the primary failure mechanism for all panels. A macro crack traversed the width of each test specimen during structural failure within the gage section. Excellent repeatability and consistency were observed for all stiffened panels. The experimental results were the motivation for developing a computational multiscale framework where a mesoscale triaxially braided composite architecture-based model is used for the material. Details of the computational multiscaling framework and results compared to the experimental study are given in Kosztowny and Waas (2021), which is part two of this two-part paper.

Application of Hyperspectral Imaging and Deep Learning for Robust Prediction of Sugar and pH Levels in Wine Grape Berries.

Remote sensing technology, such as hyperspectral imaging, in combination with machine learning algorithms, has emerged as a viable tool for rapid and nondestructive assessment of wine grape ripeness. However, the differences in terroir, together with the climatic variations and the variability exhibited by different grape varieties, have a considerable impact on the grape ripening stages within a vintage and between vintages and, consequently, on the robustness of the predictive models. To address this challenge, we present a novel one-dimensional convolutional neural network architecture-based model for the prediction of sugar content and pH, using reflectance hyperspectral data from different vintages. We aimed to evaluate the model’s generalization capacity for different varieties and for a different vintage not employed in the training process, using independent test sets. A transfer learning mechanism, based on the proposed convolutional neural network, was also used to evaluate improvements in the model’s generalization. Overall, the results for generalization ability showed a very good performance with RMSEP values of 1.118 °Brix and 1.085 °Brix for sugar content and 0.199 and 0.183 for pH, for test sets using different varieties and a different vintage, respectively, improving and updating the current state of the art.

A Deep Learning Model for Predicting DNA N6-Methyladenine (6mA) Sites in Eukaryotes

DNA N6-methyladenine (6mA) is an epigenetic modification, which is involved in many biological regulation processes like DNA replication, DNA repair, transcription, and gene expression regulation. The widespread presence of this 6mA modification in eukaryotes has been unclear until recently. Studying the genome-wide distribution of 6mA can provide a deeper understanding of the epigenetic modification process and the biological processes it involves. Existing experimental techniques are time-consuming and computational machine learning methods have room for performance improvement. DNA N6-methyladenine prediction in eukaryotic cross-species shows low performance. Hence, there is a need for a more accurate, time-efficient method to predict the distribution of 6mA sites in eukaryotes . Since deep learning architectures have shown higher accuracy, we develop a customized VGG16 architecture-based model named 6mAVGG using convolution neural networks for the prediction of DNA 6mA sites in eukaryotes. We introduce a novel 3-dimensional encoding mechanism extending the one-hot encoding method to support the input of the VGG16 model. Specifically, the 10-fold cross-validation on the benchmark datasets for the proposed model achieves higher accuracies of 98.01%, 97.44%, 99.56% respectively for cross-species, Rice, and M. musculus genomes. The proposed model outperforms existing tools for the prediction of 6mA sites and has enhanced accuracies by 2.88%, 4.2%, 0.9% respectively for cross-species, Rice , and M. musculus genomes compared to the state of the art method SNNRice6mA. The model trained with benchmark data predicts 6mA sites of other species ArabidopsisThaliana , RosaChinensis , Drosophila , and Yeast with prediction accuracy over 70%. Thus, this model can be used for the genome-wide prediction of 6mA sites in eukaryotes .

Deep Learning based Student Emotion Recognition from Facial Expressions in Classrooms

Classroom teaching assessments are intended to give valuable advice on the teaching-learning process as it happens. The finest schoolroom assessments furthermore assist as substantial foundations of information for teachers, serving them to recognize what they imparted fittingly and how they can improve their lecture content to keep the students attentive. In this paper, we have surveyed some of the recent paper works done on facial emotion recognition of students in a classroom arrangement and have proposed our deep learning approach to analyze emotions with improved emotion classification results and offers an optimized feedback to the instructor. A deep learning-based convolution neural network algorithm will be used in this paper to train FER2013 facial emotion images database and use transfer learning technique to pre-train the VGG16 architecture-based model with Cohn-Kanade (CK+) facial image database, with its own weights and basis. A trained model will capture the live steaming of students by using a high-resolution digital video camera that faces towards the students, capturing their live emotions through facial expressions, and classifying the emotions as sad, happy, neutral, angry, disgust, surprise, and fear, that can offer us an insight into the class group emotion that is reflective of the mood among the students in the classroom. This experimental approach can be used for video conferences, online classes etc. This proposition can improve the accuracy of emotion recognition and facilitate faster learning. We have presented the research methodologies and the achieved results on student emotions in a classroom atmosphere and have proposed an improved CNN model based on transfer learning that can suggestively improve the emotions classification accuracy.

Reliability and Performance Analysis of Architecture-Based Software Implementing Restarts and Retries Subject to Correlated Component Failures

High reliability and performance are essential attributes of software systems designed for critical real-time applications. To improve the reliability and performance of software, many systems incorporate some form of fault recovery mechanism. However, contemporary models of software reliability and performance rarely consider these fault recovery mechanisms. Another notable shortcoming of many software models is that they make the simplifying assumption that component failures are statistically independent, which disagrees with several experimental studies that have shown that the failures of software components can exhibit correlation. This paper presents an architecture-based model of software reliability and performance that explicitly considers a two-stage fault recovery mechanism implementing component restarts and application-level retries. The application architecture is characterized by a Discrete Time Markov Chain (DTMC) to represent the dynamic branching behavior of control between the components of the application. Correlations between the component failures are computed with an efficient numerical algorithm for a multivariate Bernoulli (MVB) distribution. We illustrate the utility of the model through a case study of an embedded software application. The results suggest that the model can be used to quantify the impact of software fault recovery and correlated component failures on application reliability and performance.

An Architecture-Based Multi-Objective Optimization Approach to Testing Resource Allocation

Software systems are widely employed in society. With a limited amount of testing resource available, testing resource allocation among components of a software system becomes an important issue. Most existing research on the testing resource allocation problem takes a single-objective optimization approach, which may not adequately address all the concerns in the decision-making process. In this paper, an architecture-based multi-objective optimization approach to testing resource allocation is proposed. An architecture-based model is used for system reliability assessment, which has the advantage of explicitly considering system architecture over the reliability block diagram (RBD)-based models, and has good flexibility to different architectural alternatives and component changes. A system cost modeling approach which is based on well-developed software cost models is proposed, which would be a more flexible, suitable approach to the cost modeling of software than the approach adopted by others which is based on an empirical cost model. A multi-objective optimization model is developed for the testing resource allocation problem, in which the three major concerns in the testing resource allocation problem, i.e., system reliability, system cost, and the total amount of testing resource consumed, are taken into consideration. A multi-objective evolutionary algorithm (MOEA), called multi-objective differential evolution based on weighted normalized sum (WNS-MODE), is developed. Experimental studies are presented, and the experiments show several results. 1) The proposed architecture-based multi-objective optimization approach can identify the testing resource allocation strategy which has a good trade-off among optimization objectives. 2) The developed WNS-MODE is better than the MOEA developed in recent research, called HaD-MOEA, in terms of both solution quality and computational efficiency. 3) The WNS-MODE seems quite robust from the sensitivity analysis results.

열차제어시스템의 아키텍처 기반 고가용도 모델 적용에 관한 연구

본 연구는 Five-9's 가용도를 달성하기 위한 고가용도 시스템의 가용도 모델을 기술한다. 현대의 철도시스템은 지속적으로 서비스가 제공되어야 하는 고가용도 시스템으로 이러한 고가용도 서비스의 중요한 특성은 이동통신, 철도시스템, 정보운영 및 웹기반 사업 등 많은 현대의 사업에 필수적이며, 아키텍처 기반의 시스템 가용도 모델은 고가용도 목표의 만족여부를 평가하는 데 매우 유용하다. 마코브 모델 접근은 관련 시스템엔지니어가 시스템고장 및 고장복구 프로세스를 모델링할 때 직관적으로 적용가능하다. 본 연구는 UML2.0을 사용하여 개선된 가용도 모델을 제시하였다. 이러한 아키텍처기반의 시스템 가용도 모델은 철도시스템에 매우 의미있게 적용될 수 있다. This work describes an availability model of highly available systems to achieve Five-9's availability. Modern railway systems have raised users' expectations of powerful "always on" services. The crucial characteristics of these highly available services are essential to many modern businesses area, such as telecommunications, railway systems, information operations, Web-based businesses, and so on. The architecture-based model of system availability is useful to assess the feasibility of meeting a high availability target. The Markov model approach is straightforward for relative system engineers to adapt when they model highly available system failure and the failure recovery process. This work proposed the improved availability model through UML2.0. It is shown that the architecture-based model of system availability is a good reasonable by its application of the railway systems.

A Dynamic Service Pool Size Configuration Mechanism for Service-Oriented Workflow

Service-oriented workflows are the fundamental structures in service-oriented applications and changes in the workflow could cause dramatic changes in system reliability. In several ways to re-heal workflows in execution, re-sizing service pools in the workflow is practical and easy to implement. In order to quickly adjust to workflow or environmental changes, this paper presents a dynamic service pool size configuration mechanism from the point of view of maintaining workflow reliability. An architecture-based reliability model is used to evaluate the overall reliability of a workflow with service pools and an optimal method is proposed to get the combination of service pool size aiming at minimizing the sum of service pool size subject to the workflow reliability requirement. A case study is used to explain this method and experiment results show how to change service pool size to meet the workflow reliability requirements.