Intelligent decision-making (IDM) is a cornerstone of artificial intelligence (AI) designed to automate or augment decision processes. Modern IDM paradigms integrate advanced frameworks to enable intelligent agents to...
详细信息
Recent years have witnessed great progress in deep learning based object detection. However, due to the domain shift problem, applying off-the-shelf detectors to an unseen domain leads to significant performance drop....
详细信息
ISBN:
(数字)9781728171685
ISBN:
(纸本)9781728171692
Recent years have witnessed great progress in deep learning based object detection. However, due to the domain shift problem, applying off-the-shelf detectors to an unseen domain leads to significant performance drop. To address such an issue, this paper proposes a novel coarse-to-fine feature adaptation approach to cross-domain object detection. At the coarse-grained stage, different from the rough image-level or instance-level feature alignment used in the literature, foreground regions are extracted by adopting the attention mechanism, and aligned according to their marginal distributions via multi-layer adversarial learning in the common feature space. At the fine-grained stage, we conduct conditional distribution alignment of foregrounds by minimizing the distance of global prototypes with the same category but from different domains. Thanks to this coarse-to-fine feature adaptation, domain knowledge in foreground regions can be effectively transferred. Extensive experiments are carried out in various cross-domain detection scenarios. The results are state-of-the-art, which demonstrate the broad applicability and effectiveness of the proposed approach.
data partition and replication mechanisms directly determine query execution patterns in parallel database systems, which have a great impact on system performance. Recently, there have been some workload-aware data s...
详细信息
In spite that Federated Learning (FL) is well known for its privacy protection when training machine learning models among distributed clients collaboratively, recent studies have pointed out that the naive FL is susc...
详细信息
This brief explores the approximation properties of a unique basis expansion based on Pascal's triangle, which realizes a sampled-data driven approach between a continuous-time signal and its discrete-time represe...
详细信息
ISBN:
(数字)9781728158556
ISBN:
(纸本)9781728158563
This brief explores the approximation properties of a unique basis expansion based on Pascal's triangle, which realizes a sampled-data driven approach between a continuous-time signal and its discrete-time representation. The roles of certain parameters, such as sampling time interval or model order, and signal characteristics, i.e., its curvature, on the approximation are investigated. Approximate errors in one and multiple-step predictions are analyzed. Furthermore, time-variant approximations under the thresholds of signal curvature are employed to narrow errors and provide flexibilities.
Breast cancer is one of the most common causes of death among women worldwide. Early detection helps in reducing the number of deaths. Automated 3D Breast Ultrasound (ABUS) is a newer approach for breast screening, wh...
详细信息
With the rapid development of Internet of Things (IoT) technology, completing a transaction with intelligent devices has been an essential communication mode in our daily life. Generally, lightweight smart devices con...
详细信息
Exploring GUIs of Android apps plays a key role in many important scenarios such as functional testing (e.g., finding crash errors), security analysis (e.g., identifying malicious behav-iors) and competitive analysis ...
详细信息
ISBN:
(数字)9781728156194
ISBN:
(纸本)9781728156200
Exploring GUIs of Android apps plays a key role in many important scenarios such as functional testing (e.g., finding crash errors), security analysis (e.g., identifying malicious behav-iors) and competitive analysis (e.g., storyboarding app features). To automate GUI exploration, existing techniques often try to visit as many GUI pages as possible via specific strategies, e.g., random (like Monkey) or heuristic (like Stoat, A 3 E). However, their effectiveness is still unclear and much under-explored. To this end, we conducted the first study in this paper to understand and characterize their limitations by carefully analyzing the coverage reports from a set of real-world, open-source apps. Through this study, we identified three key limitations due to the lack of dependency knowledge during exploration, i.e., widget-page dependency, widget-widget dependency and system-event dependency. To overcome them, we introduce dependency-informed exploration, an automated approach that leverages static dependency analysis to effectively improve GUI exploration performance. Given an app, our approach first constructs a GUI page transition model that captures the dependencies between GUI widgets, and then guides GUI exploration during a depth-first traversal. We realized our approach as a tool named Gesda, and evaluated it on 70 open-source Android apps. The results show Gesda outperforms existing state-of-the-art GUI exploration techniques, i.e., Monkey and Stoat. Additionally, Gesda uncovers 4 previously unknown crashes in 4 apps as a by-product of GUI exploration due to the benefit of dependency knowledge, while Monkey and Stoat have not discovered them.
A latent factor analysis (LFA) model can efficiently address a high-dimensional and sparse (HiDS) matrix with a stochastic gradient descent (SGD) algorithm. However, an SGD-based LFA model's performance depends he...
详细信息
ISBN:
(纸本)9781665426220
A latent factor analysis (LFA) model can efficiently address a high-dimensional and sparse (HiDS) matrix with a stochastic gradient descent (SGD) algorithm. However, an SGD-based LFA model's performance depends heavily on its hyper-parameters. The popular method is based on grid-search, which fails because of expensive computation and time-consuming. Aiming at implementing a hyper-parameter-free LFA model, this study proposes an adaptive moment estimation-incorporated particle swarm optimization (Adam-PSO) algorithm that efficiently addresses the premature issues in a PSO algorithm. With achieved the hyper-parameter adaptation in an SGD-based LFA model, an Adam-PSO-based LFA (APL) model possessing hyper-parameter-free training is further implemented. Empirical studies on four HiDS matrices indicate that compared with state-of-the-art models with hyper-parameter adaptation settings, an APL model achieves the most efficient hyper-parameter-free training and highly competitive prediction accuracy for missing data of an HiDS matrix. Hence, it fits the need of real applications with high scalability and efficiency.
Recent years have witnessed great progress in deep learning based object detection. However, due to the domain shift problem, applying off-the-shelf detectors to an unseen domain leads to significant performance drop....
详细信息
暂无评论