There is, I think, something ethereal about i —the square root of minus one. I remember first hearing about it at school. It seemed an odd beast at that time—an intruder hovering on the edge of reality.

Usually familiarity dulls this sense of the bizarre, but in the case of i it was the reverse: over the years the sense of its surreal nature intensified. It seemed that it was impossible to write mathematics that described the real world in …

I and i

Matrix Factorization Techniques for Recommender Systems

/pdf/a-break-in-the-clouds-towards-a-cloud-definition-svovsw1r5a.pdf

A Break in the Clouds: Towards a Cloud Definition

Network virtualization is recognized as an enabling technology for the future Internet. It aims to overcome the resistance of the current Internet to architectural change. Application of this technology relies on algorithms that can instantiate virtualized networks on a substrate infrastructure, optimizing the layout for service-relevant metrics. This class of algorithms is commonly known as "Virtual Network Embedding (VNE)" algorithms. This paper presents a survey of current research in the VNE area. Based upon a novel classification scheme for VNE algorithms a taxonomy of current approaches to the VNE problem is provided and opportunities for further research are discussed.

Virtual Network Embedding: A Survey

In the past five years there has been a dramatic increase in work on Search-Based Software Engineering (SBSE), an approach to Software Engineering (SE) in which Search-Based Optimization (SBO) algorithms are used to address problems in SE. SBSE has been applied to problems throughout the SE lifecycle, from requirements and project planning to maintenance and reengineering. The approach is attractive because it offers a suite of adaptive automated and semiautomated solutions in situations typified by large complex problem spaces with multiple competing and conflicting objectives.This article1 provides a review and classification of literature on SBSE. The work identifies research trends and relationships between the techniques applied and the applications to which they have been applied and highlights gaps in the literature and avenues for further research.

/pdf/search-based-software-engineering-trends-techniques-and-3b0bwaagfi.pdf

Search-based software engineering: Trends, techniques and applications

This article puts forward a cross-AS (Autonomous System, AS) IP trackback method, called FFDPM, for tracing DDoS attack. FFDPM exploits the BGP AS-PATH attribute, which indicates the AS sequence one packet needs to pass through before reaching its destination, to make the Border Gateway (BG) divide the AS Number (ASN) in a flexible manner based on the distance between itself and the packet's aim AS, so as to improve the applicable scope and reduce the number of packets needed for path reconstruction. In order to resist the forging mark information, this method also proposes a mark information authentication scheme, which stores the authentication information subtly in the idle room that has not been used by mark information. The theoretical analysis shows that FFDPM can trace DDoS attacks effectively.

A cross-AS traceback method based on flexible fragmentation of path information

Relation extraction (RE), an important information extraction task, faced the great challenge brought by limited annotation data. To this end, distant supervision was proposed to automatically label RE data, and thus largely increased the number of annotated instances. Unfortunately, lots of noise relation annotations brought by automatic labeling become a new obstacle. Some recent studies have shown that the teacher-student framework of knowledge distillation can alleviate the interference of noise relation annotations via label softening. Nevertheless, we find that they still suffer from two problems: propagation of inaccurate dark knowledge and constraint of a unified distillation temperature. In this article, we propose a simple and effective Multi-instance Dynamic Temperature Distillation (MiDTD) framework, which is model-agnostic and mainly involves two modules: multi-instance target fusion (MiTF) and dynamic temperature regulation (DTR). MiTF combines the teacher’s predictions for multiple sentences with the same entity pair to amend the inaccurate dark knowledge in each student’s target. DTR allocates alterable distillation temperatures to different training instances to enable the softness of most student’s targets to be regulated to a moderate range. In experiments, we construct three concrete MiDTD instantiations with BERT, PCNN, and BiLSTM-based RE models, and the distilled students significantly outperform their teachers and the state-of-the-art (SOTA) methods.

MiDTD: A Simple and Effective Distillation Framework for Distantly Supervised Relation Extraction

With the emergence and rapid development of event-based social networks (EBSNs), the problem of how to maximize the number of participants for social events has been widely studied. Most previous studies on this problem focus on a single social event. However, in real-world scenarios, there are multiple social events and they may compete for influential users to maximize the number of participants since they may be held during the same time period and in acceptable geographic locations. To bridge this gap, in this paper, we propose to study a novel problem of Fair-Aware Competitive Event Influence Maximization (FCE-IM) . This problem is how to select influential users for each social event on the premise of fairness to maximize the number of participants per event in the competitive scenario. To solve this problem, we first propose a propagation model, named E-LT , to describe the information propagation process in competitive situation in EBSNs. Then, we propose a Randomized Algorithm based on Cross Entropy method , named RACE . It can optimally assign node selection probability for each node and thus effectively approach to the maxima. We conduct extensive experiments using two real-world datasets and experimental results demonstrate the efficiency and effectiveness of our algorithm.

Fair-Aware Competitive Event Influence Maximization in Social Networks

Network virtualization has been proposed as a promising way for running multiple customized virtual networks (VNs) on a shared infrastructure. However, how to provide reliable VN against substrate infrastructure failures has become an increasingly important issue. Meanwhile the substrate network resource cost should be minimized under VN reliability guarantees to maximize the revenue for the Infrastructure Providers (InPs). In this paper, we present a novel heuristic VN mapping algorithm called RVNM, which includes a node mapping scheme based on substrate node reliability awareness and a link mapping scheme based on shared-path protection. Simulation results show that algorithm can improve VN robustness without reserving protection resource and gain near optimal network bandwidth usage compared to the previous algorithms.

Virtual Network Mapping Algorithm with Substrate Node Reliability Awareness and Shared-Path Protection

The autonomous behaviors of peers are usually neglected in the topology formation and adaptation in P2P network, thus a large number of Free-rider and Malicious peers would reside in the center of network, which degrades the user perceived QoE and overall network performance. In this paper, a novel and fully distributed approach Topology Adaptation algorithm based on Reciprocal Contribution (TARC) is proposed to address such problem. In our paradigm, the topology would be periodical optimized according to Reciprocal Contribution Capacity (RCC) of peers. The RCC reflects the historical behavior of a peer with regard to its content provision and transmission capability as well as nodes locating capability. Peer that possesses higher RCC, namely contributor, as it were more cooperative would be easily and likely to establish connections with others. On the other hand, malicious and free-riding nodes would be recognized and forced to the marginal of P2P network. It is shown through experiments that TARC effectively suppress the fake services and prevent non-contribution peer from occupying the center of network. What’s more, there is a significant improvement in term of overall search quality and downloading time, for contributing peers in a network that applies our algorithm.

Sen Su

Papers

A cross-AS traceback method based on flexible fragmentation of path information

MiDTD: A Simple and Effective Distillation Framework for Distantly Supervised Relation Extraction

Fair-Aware Competitive Event Influence Maximization in Social Networks

Virtual Network Mapping Algorithm with Substrate Node Reliability Awareness and Shared-Path Protection

TARC: A Novel Topology Adaptation Algorithm Based on Reciprocal Contribution in Unstructured P2P Networks