Sign boundary and hand articulation feature recognition in Sign Language videos

Machine Translation - In this paper we present a recommendation system for (semi-)automatic annotation of sign language videos exploiting deep learning techniques, which handle handshape...     

Adaptive Smooth Simulcast Protocol (ASSP) for Video Applications: Description and Performance Evaluation

In this paper, we present Adaptive Smooth Simulcast Protocol (ASSP) for simulcast transmission of multimedia data over best-effort networks. ASSP is a new multiple-rate protocol that implements a single rate TCP-friendly protocol as the underlying congestion control mechanism for each simulcast stream. The key attributes of ASSP are: (a) TCP-friendly behavior, (b) adaptive per-stream transmission rates, (c) adaptive scalability to large sets of receivers and (d) smooth transmission rates that are suitable for multimedia applications. We evaluate the performance of ASSP under an integrated simulation environment which combines the measurements of both network and video performance metrics. We also compare ASSP against other proposed solutions and the results demonstrate that the performance of ASSP is significantly better than the tested solutions. Finally, ASSP is a practical solution with very low implementation complexity for video transmission over best-effort networks.     

Multiscale quality assessment of Global Human Settlement Layer scenes against reference data using statistical learning

A method for quality assessment of the Global Human Settlement Layer scenes against reference data is presented. It relies on two settlement metrics; the local average and gradient functions that quantify the notions of settlement density and flexible settlement limits respectively. They are both utilized as generalization functions for increasing the level of abstraction of the sets under comparison. Generalization compensates for inaccuracies of the automatic target extraction method and can be computed at multiple scales. The comparison between the target built-up layers and the reference data employs an ordered multi-scale, linear regression computing the goodness of fit measure R²$R^2$. An optimized assessment procedure is investigated in a pilot study and is further employed in a big data exercise. A newly introduced quality metric returns the agreement between automatically extracted built-up from a set of 13605 scenes and the MODIS 500 urban layer, that was found too be as high as 91% for selected sensors. A final experiment attempts a performance increase at lower scales by correlating the target layer with automatically selected training subsets. At 50 m the adjusted R²$R^2$ increases by 3% with a mean squared error improvement of 2% compared to the performance achieved without statistical learning. The experiment suggests that the GHSL assessment at a global scale can be carried out based on limited high resolution reference data of minimal spatial coverage.     

A novel HMM-based learning framework for improving dynamic wireless push system performance

A new machine learning framework is introduced in this paper, based on the hidden Markov model (HMM), designed to provide scheduling in dynamic wireless push systems. In realistic wireless systems, the clients’ intentions change dynamically; hence a cognitive scheduling scheme is needed to estimate the desirability of the connected clients. The proposed scheduling scheme is enhanced with self-organized HMMs, supporting the network with an estimated expectation of the clients’ intentions, since the system’s environment characteristics alter dynamically and the base station (server side) has no a priori knowledge of such changes. Compared to the original pure scheme, the proposed machine learning framework succeeds in predicting the clients’ information desires and overcomes the limitation of the original static scheme, in terms of mean delay and system efficiency.     

Sequentially optimal repeated coalition formation under uncertainty

Coalition formation is a central problem in multiagent systems research, but most models assume common knowledge of agent types. In practice, however, agents are often unsure of the types or capabilities of their potential partners, but gain information about these capabilities through repeated interaction. In this paper, we propose a novel Bayesian, model-based reinforcement learning framework for this problem, assuming that coalitions are formed (and tasks undertaken) repeatedly. Our model allows agents to refine their beliefs about the types of others as they interact within a coalition. The model also allows agents to make explicit tradeoffs between exploration (forming “new” coalitions to learn more about the types of new potential partners) and exploitation (relying on partners about which more is known), using value of information to define optimal exploration policies. Our framework effectively integrates decision making during repeated coalition formation under type uncertainty with Bayesian reinforcement learning techniques. Specifically, we present several learning algorithms to approximate the optimal Bayesian solution to the repeated coalition formation and type-learning problem, providing tractable means to ensure good sequential performance. We evaluate our algorithms in a variety of settings, showing that one method in particular exhibits consistently good performance in practice. We also demonstrate the ability of our model to facilitate knowledge transfer across different dynamic tasks.     

Minimizing information disclosure to third parties in social login platforms

Over the past few years, a large and ever increasing number of Web sites have incorporated one or more social login platforms and have encouraged users to log in with their Facebook, Twitter, Google, or other social networking identities. Research results suggest that more than two million Web sites have already adopted Facebook’s social login platform, and the number is increasing sharply. Although one might theoretically refrain from such social login features and cross-site interactions, usage statistics show that more than 250 million people might not fully realize the privacy implications of opting-in. To make matters worse, certain Web sites do not offer even the minimum of their functionality unless users meet their demands for information and social interaction. At the same time, in a large number of cases, it is unclear why these sites require all that personal information for their purposes. In this paper, we mitigate this problem by designing and developing a framework for minimum information disclosure in social login interactions with third-party sites. Our example case is Facebook, which combines a very popular single sign-on platform with information-rich social networking profiles. Whenever users want to browse to a Web site that requires authentication or social interaction using a Facebook identity, our system employs, by default, a Facebook session that reveals the minimum amount of information necessary. Users have the option to explicitly elevate that Facebook session in a manner that reveals more or all of the information tied to their social identity. This enables users to disclose the minimum possible amount of personal information during their browsing experience on third-party Web sites.     

Hyperconnected attribute filters based on k-flat zones

In this paper, we present a new method for attribute filtering, combining contrast and structural information. Using hyperconnectivity based on k-flat zones, we improve the ability of attribute filters to retain internal details in detected objects. Simultaneously, we improve the suppression of small, unwanted detail in the background. We extend the theory of attribute filters to hyperconnectivity and provide a fast algorithm to implement the new method. The new version is only marginally slower than the standard Max-Tree algorithm for connected attribute filters, and linear in the number of pixels or voxels. It is two orders of magnitude faster than anisotropic diffusion. The method is implemented in the form of a filtering rule suitable for handling both increasing (size) and nonincreasing (shape) attributes. We test this new framework on nonincreasing shape filters on both 2D images from astronomy, document processing, and microscopy, and 3D CT scans, and show increased robustness to noise while maintaining the advantages of previous methods.     

DoS attacks exploiting signaling in UMTS and IMS

The Universal Mobile Telecommunication Standard (UMTS) is continuously evolving to meet the growing demand of modern mobile and Internet applications for high capacity and advanced features in security and quality of service. Although admittedly enhanced in terms of security when compared to 2G systems, UMTS still has weaknesses that can lead to security incidents. In this paper, we investigate the vulnerabilities of the UMTS security architecture that can be exploited by a malicious individual to mount Denial of Service (DoS) attacks. Our focus is on signaling-oriented attacks above the physical layer. We describe and analyze several novel attacks that can be triggered against both core UMTS architecture as well as hybrid UMTS/WLAN realms. An additional contribution of this paper is the presentation of an extensive survey of similar attacks in UMTS and related protocol infrastructures such as IP Multimedia Subsystem (IMS). Finally, we offer some suggestions that would provide greater tolerance to the system against DoS attacks.     

手动式电磁定位及图像导航的短径癌症放射治疗系统

提出一个用于高剂量短径癌症放射治疗的手动式电磁定们及图像导航系统，该系统主要由4个模块组成：预处理、术前规划、术内导航和术后计划，它们分别对应高剂量短径癌症放射治疗的4个流程：在术前规划阶段，用户可交互地设计探针放置位置以及三维重建病灶轮廓；在手术进行过程中，六自由度的电磁式定位技术实时而精确地跟踪探针位置，高质量的可视化技术指引医生完成手术探针放置任务；手术探针全部放置完毕后，术后计划模板计算剂量分布并利用后装机设备进行放射治疗。初步实验表明，计算机辅助的探针定位比手工定位精度提高2-3倍，手术时间减少5倍以上，同时也减轻了病人所承受的放射剂量。     

Named-Entity Recognition from Greek and English Texts

Named-entity recognition (NER) involves the identification and classification of named entities in text. This is an important subtask in most language engineering applications, in particular information extraction, where different types of named entity are associated with specific roles in events. In this paper, we present a prototype NER system for Greek texts that we developed based on a NER system for English. Both systems are evaluated on corpora of the same domain and of similar size. The time-consuming process for the construction and update of domain-specific resources in both systems led us to examine a machine learning method for the automatic construction of such resources for a particular application in a specific language.