While neural networks can be approximated by linear models as their width increases, certain properties of wide neural networks cannot be captured by linear models. In this work we show that recently proposed Neural Q...
As communities represent similar opinions, similar functions, similar purposes, etc., community detection is an important and extremely useful tool in both scientific inquiry and data analytics. However, the classic m...
详细信息
In this paper we introduce a new approach to characterizing image quality: visual equivalence, Images are visually equivalent if they convey the same information about object appearance even if they are visibly differ...
详细信息
ISBN:
(纸本)9780892082810
In this paper we introduce a new approach to characterizing image quality: visual equivalence, Images are visually equivalent if they convey the same information about object appearance even if they are visibly different. In a series of psychophysical experiments we explore how object geometry, material, and illumination interact to produce images that are visually equivalent, and we identify how two kinds of transformations on illumination fields (blurring and warping) influence observers' judgments of equivalence. We use the results of the experiments to derive metrics that can serve as visual equivalence predictors (VEPs) and we generalize these metrics so they can be applied to novel objects and scenes. Finally we validate the predictors in a confirmatory study, and show that they reliably predict observer's judgments of equivalence. Visual equivalence is a significant new approach to measuring image quality that goes beyond existing visible difference metrics by leveraging the fact that some kinds of image differences do not matter to human observers. By taking advantage of higher order aspects of visual object coding, visual equivalence metrics should enable the development of powerful new classes of image capture, compression, rendering, and display algorithms.
Researchers emphasize the importance of hardware accelerators for mathematical morphology. If there are any issues, the hardware architecture may need to be redesigned. Thus, we propose a novel, reconfigurable hardwar...
详细信息
With support from HBCU-UP (Historically Black College and University Undergraduate program) at National science Foundation (NSF), the ACTION (Advanced Curriculum and Technology-Based Instructional Opportunities Networ...
详细信息
This paper proposes a framework called GHVC-Net that uses the graph neural network (GNN) model to approximate each solution's hypervolume contribution (HVC). GHVC-Net is permutation invariant and can handle soluti...
详细信息
We study transfer learning for estimation in latent variable network models. In our setting, the conditional edge probability matrices given the latent variables are represented by P for the source and Q for the targe...
The importance of weak social ties in professional communities is well studied and widely accepted. In our paper we analyze the structure of strong ties based on the co-authorship relation and use the formal concept a...
详细信息
The importance of weak social ties in professional communities is well studied and widely accepted. In our paper we analyze the structure of strong ties based on the co-authorship relation and use the formal concept analysis framework to figure out weak ties. The research is motivated by fast growing need in cross-disciplinary research, which requires experts from different areas to understand the bigger picture and identify potential fellows for collaborative research projects in nearest future.
Pulmonary nodules are small, round, or oval-shaped growths on the lungs. They can be benign (noncancerous) or malignant (cancerous). The size of a nodule can range from a few millimeters to a few centimeters in diamet...
详细信息
Pulmonary nodules are small, round, or oval-shaped growths on the lungs. They can be benign (noncancerous) or malignant (cancerous). The size of a nodule can range from a few millimeters to a few centimeters in diameter. Nodules may be found during a chest X-ray or other imaging test for an unrelated health problem. In the proposed methodology pulmonary nodules can be classified into three stages. Firstly, a 2D histogram thresholding technique is used to identify volume segmentation. An ant colony optimization algorithm is used to determine the optimal threshold value. Secondly, geometrical features such as lines, arcs, extended arcs, and ellipses are used to detect oval shapes. Thirdly, Histogram Oriented Surface Normal Vector (HOSNV) feature descriptors can be used to identify nodules of different sizes and shapes by using a scaled and rotation-invariant texture description. Smart nodule classification was performed with the XGBoost classifier. The results are tested and validated using the Lung Image Consortium database (LICD). The proposed method has a sensitivity of 98.49% for nodules sized 3–30 mm.
暂无评论