Abstract Convolution neural networks have been widely used in the field of computer vision, which effectively solve practical problems. However, the loss function with fixed parameters will affect the training efficiency and even lead to poor prediction accuracy. In particular, when there is a class imbalance in the data, the final result tends to favor the large‐class. In detection and recognition problems, the large‐class will dominate due to its quantitative advantage, and the features of few‐class can be not fully learned. In order to learn few‐class, batch nuclear‐norm maximization is introduced to the deep neural networks, and the mechanism of the adaptive composite loss function is established to increase the diversity of the network and thus improve the accuracy of prediction. The proposed loss function is added to the crowd counting, and verified on ShanghaiTech and UCF_CC_50 datasets. Experimental results show that the proposed loss function improves the prediction accuracy and convergence speed of deep neural networks.

Most colorectal cancers are caused by colorectal adenomatous polyps, early screening for colonic polyps is of great clinical importance. However, polyps have diverse appearance, such as shape and size; polyps are difficult to distinguish from mucosa. To address these problems, this paper proposes a boundary attention and negative matrix L1 regularization loss function synergistic segmentation method. Firstly, a branch of boundary attention is added to the last detail finding branch of PraNet [16]. Secondly, the inner polyp region of the boundary is complemented using negative matrix L1 regularization iterations. The synergy of the two terms can refine object areas, which will improve the segmentation accuracy. We conducted extensive experimental evaluations on four publicly available datasets, results show that our method is superior to other models. Especially on CVC-ClinicDB dataset, compared with PraNet, Dice is improved by 4.3% and IoU is improved by 5.9%.

Network theory has provided a new analytical tool for the study of human trajectory and has also achieved rapid development in the complex network field. Conventional network model or complex network model ignores some details and cannot display the most remarkable features for a GPS based personal trajectory. It is necessary to set up a new personal trajectory model. For the purpose of researching the characteristics of trajectory for one person in a long time, we collected a GPS based personal LifeLog dataset named Liu Lifelog in the past 9 years. This paper analyzed the Liu Lifelog and proposed a ring structure personal trajectory (RSPT) model based on the basic complex network model. We discussed the definition, source, characteristic and attribute of the RSPT model and tested the model with the dataset which was provided by the Geolife project and verified that the model described the characteristic of trajectory for a person well. The result shows that this model is feasible and it can predict the human behavior characteristics more accurately and effectively.

In the underdetermined model , where is a K ‐group sparse matrix (i.e. it has no more than K non‐zero rows), the matrix may be also perturbed. Theoretically, a more relaxed condition means that fewer measurements are required to ensure sparse recovery. In this study, a relaxed sufficient condition is proposed for greedy block coordinate descent (GBCD) under total perturbations based on the restricted isometry property in order to guarantee that the support of is recovered. We also show that GBCD fails in a more general case when .

Researchers utilised various types of information in active contour models to define new energy functionals for image segmentation. These models aim to extract all potential objects from the background, but non‐target objects and noise are also obtained. In this study, the authors aim to extract target objects with sparse representation method. The original indicator function (a binary function) with respect to the level set function is used to represent the foreground (value is 1) and background(value is 0). From another point of view, an indicator function can be represented by linear combination of a set of the basis function. Firstly, by a label operator for the indicator function in each iteration, every connected area is represented by a basis function. Secondly, the linear combination of these basis functions is used to represent objects. Finally, through the sparsity constraint of coefficients of basis functions, the object extraction is viewed as a sparse representation problem. Meanwhile, a corresponding improved orthogonal matching pursuit algorithm is designed to obtain the ideal results. Experiments demonstrate that the proposed method has superior performance in object extraction compared with state‐of‐the‐art active contour models. Furthermore, the proposed method also increases the flexibility of applications.

Practically, sparsity is expressed not in terms of an orthonormal basis but in terms an overcomplete dictionary. There are many practical examples in which a signal of interest is sparse in an overcomplete dictionary. The authors propose a new algorithm signal space fast iterative hard thresholding (SSFIHT) for the recovery of dictionary‐sparse signals. Under total perturbations, using D ‐restricted isometry property ( D ‐RIP), the authors provide the proof of convergence for SSFIHT. Comparing with the error of oracle recovery, it is easy to see that SSFIHT can provide oracle‐order recovery performance against total perturbations. Numerical simulations are performed to verify the conclusions.

In active contour models (snakes), various vector force fields replacing the gradient of the original external energy in the equations of motion are a popular way to extract the object boundary. Gradient descent method is usually used to obtain the equations of motion by minimising the energy functional. However, it always suffers from local minimum in extracting complex geometries because of non‐convex functional. Gradient descent method with adaptive momentum term is proposed in this study. First, an acceleration function of evolution is defined. Then, the adaptive momentum term is obtained by calculating the product between the edge stopping function and the defined acceleration function. Finally, adaptive momentum is compatible with the snakes. The edge stopping function is used to decide the influence region of the momentum, whereas the defined acceleration function determines the magnitude of the momentum. It is used to extract the complex geometries (such as deep concavity) when adding the adaptive momentum into some snakes, such as gradient vector field or vector field convolution snakes. On the other hand, the proposed method also accelerates the rate of convergence. It can be applied to extract a single object in real images. The experimental results show that the proposed method is effective and efficient.

Nowadays, SaaS is increasingly used by web-based applications for the benefits and profits it brings to both users and service providers. It is significative if service providers can automatically convert traditional applications into SaaS mode, a SaaSify tool is needed urgently. In this paper, we analyze and conclude the new challenges of automatically SaaSify web-based application, propose several key technologies for SaaSifying, and further propose SaaSify Flow Language (SFL) to model and implement SaaSify process, finally, we use a case study to show the effects of proposed tool, and the performance experiments prove that the proposed approach is efficient and effective.

Software as a service (SaaS) is an emerging software framework in which business data and logic typically integrate with other applications. It requires a unified subscriber to describe SaaS to make for easy integration, however, SaaS provides services to different tenants by running only one instance. In order to satisfy personalized needs from different tenants, the business logic becomes correspondingly complex. As this logic is cumbersome to reveal to every individual tenant, we propose the use of Web Services Conversation Language (WSCL) to express the views of tenant and provider separately. To overcome deficiencies in WCSL for expressing heterogeneous data, process rules, and business rules, we extend the syntax of WSCL. We also put forward a new modeling method for constructing SaaS Service, describing the modeling process and the algorithm for obtaining the tenant model from the business model. In conclusion, we describe the modeling tools and validation methods.