January 2017 - Dm-meeting - mailman.cs.vt.edu

New Standard ACM Conference Proceedings Template
by Sorour Ekhtiari Amiri 27 Jan '17

27 Jan '17

Hi everyone, ACM updated its template for 2017 conferences. In a case you want to submit a paper to ACM conferences such as KDD, you can find the new template from the following link: http://www.acm.org/publications/proceedings-template Best regards, Sorour

3 2

Fwd: [Bburg-fac] Revised Agenda: CS Faculty Candidate _Kyle Fox
by B. Aditya Prakash 27 Jan '17

27 Jan '17

Hi guys, I would encourage you guys to attend Kyle's talk, and also to stay after for the meeting with students. This would be useful when you yourself come on the job market and interview :-) I'll send such reminders during this season, when I feel you should attend the talk/meet with the student. cheers, - Aditya -------- Forwarded Message -------- Subject: [Bburg-fac] Revised Agenda: CS Faculty Candidate _Kyle Fox Date: Tue, 24 Jan 2017 16:30:33 -0500 From: Debbie Cole via Bburg-fac <bburg-fac(a)cs.vt.edu> Reply-To: Debbie Cole <dcole(a)cs.vt.edu> To: faculty(a)cs.vt.edu, bburg-staff(a)cs.vt.edu <bburg-staff(a)cs.vt.edu>, hdhillon(a)vt.edu, kraylux(a)vt.edu, VT <rittenfa(a)vt.edu>, Ed Nelson <enelson(a)vt.edu> *Revised agenda below and attached* -- Debbie Cole Executive Assistant to the Dept. Head Department of Computer Science 2202 Kraft Drive (0902) Suite 1106, Knowledge Works II Blacksburg, VA 24060 Phone: (540)231-8452 Fax: (540) 231-4240 Email: dcole(a)cs.vt.edu _______________________________________________ Bburg-staff mailing list Bburg-staff(a)cs.vt.edu

1 0

Group Meeting Schedule (Spring 2017)
by Sorour Ekhtiari Amiri 26 Jan '17

26 Jan '17

Hi everyone, I hope you had a wonderful break! This semester, our weekly meeting is scheduled on *Mondays, 4-5pm in Torg. 3160A*. We will have our first meeting next week, where I will present our AAAI paper. Title: "SnapNETS: Automatic Segmentation of Network Sequences with Node Labels" Location: Torg. 3160A Time: Monday, Jan 30th, 4:00 - 5:00 pm Best regards, Sorour

1 0

PASS: Congratulations to Yao for his prelim exam
by B. Aditya Prakash 25 Jan '17

25 Jan '17

Great job, Yao! - Aditya

2 1

Preliminary exam: Yao Zhang, Torgersen 3160A, 11am-1pm, Jan 24 (Tomorrow)
by Yao Zhang 24 Jan '17

24 Jan '17

Hi everyone, Yao Zhang will have his preliminary exam tomorrow (Wednesday) at 11am-1pm in Torgersen 3160A. Everyone is welcomed to attend his exam. His title, abstract and abstract are given below. Thank you. Regards, Yao ================== ***Title*** Optimizing and Understanding Network Structure for Diffusion *** Link of Yao's proposal*** http://people.cs.vt.edu/~yaozhang/prelim/thesis.pdf *** Abstract *** Given a population contact network, how to distribute vaccines to individuals to effectively control a flu epidemic? Similarly, in social media websites, how to choose the best communities/groups to stop rumors from spreading? Additionally, how to nd the best accounts that bridge celebrities and ordinary users? These questions are related to diffusion (aka propagation) phenomena. Diffusion can be treated as a behavior of spreading contagions (like viruses, ideas, memes, etc) on some underlying network. It is omnipresent in areas such as social media, public health, and cyber-security. Examples include diseases like flu spreading on person-to-person contact networks, memes disseminating by online adoption over online friendship networks, and malware propagating among computer networks. When a contagion spreads, network structure (like nodes/edges/groups, etc.) plays an important role in determining the outcome. For instance, a rumor, if propagated by celebrities, can possibly go viral. Similarly, an epidemic can die out quickly, if vulnerable demographic groups are successfully targeted for vaccination. Hence in this thesis, we aim to optimize and understand network structure better in light of diffusion. We optimize graph topologies by removing nodes/edges for controlling rumors/viruses from Hence in this thesis, we aim to optimize and understand network structure better in light of diffusion. We optimize graph topologies by removing nodes/edges for controlling rumors/viruses from spreading, and gain a deeper understanding of a network in terms of diffusion by exploring how nodes group together for similar roles of dissemination. We develop several novel graph mining algorithms, with different levels of granularity (node/edge level to group/community level), from model-driven and data-driven perspectives, focusing on topics like immunization on networks, graph summarization, and community detection. 1. Model-driven. Diffusion processes are usually described using mathematical models (e.g., the Independent Cascade (IC) model in social media, and the Susceptible-InfectiousRecovered (SIR) model in epidemiology). Given such models, we propose to optimize network structure for controlling propagation (the immunization problem) in several practical settings. We develop efficient algorithms for different interventions, such as vaccination (node removal) and quarantining (edge removal). In addition, we study the graph coarsening problem to obtain a better understanding of relations of nodes when a contagion is propagating. Our goal is to get a much smaller representation of a large network, while preserving its diffusive properties. 2. Data-driven. In contrast to model-based methods, data-driven approaches directly work on propagation data by relaxing modeling assumptions of diffusion. Hence they are more realistic for contagion related problems. This is due to the fact that typically diffusion processes are very complicated in reality, and it is challenging or even impossible to pick the most suited model to describe them. In this part of the thesis, we use data directly from sources like tweets and electronic healthcare records. We first propose a novel community detection problem to discover \bridge" and \celebrity" communities from social media data, and design case studies to understand roles of nodes/communities using diffusion. Second, we plan to develop data-driven immunization strategies to stop rumors or allocate vaccines by optimizing network topologies, using over 10 million tweets or national-level diagnostic patient data. Our work has many applications in multiple areas such as epidemiology, sociology and computer science. For example, it can help Our work has many applications in multiple areas such as epidemiology, sociology and computer science. For example, it can help discover communities of users in terms of propagations in social media. Similarly, our work on efficient immunization algorithms can guide better policies to control epidemics.

1 0

Yao's prep talk for his prelim exam (Fri Jan 20 3pm MCB133C)
by yaozhang 20 Jan '17

20 Jan '17

Hi everyone, We will start our group meeting this week. Yao will give his prep talk for his prelim exam this Friday. Please see the end for the details of his proposal. You are very welcomed to give any suggestions. TITLE: OPTIMIZING AND UNDERSTANDING NETWORK STRUCTURE FOR DIUSION TIME: FRIDAY, JAN. 20, 3:00PM - 5:00PM LOCATION: MCB133C *Note that this may be the tentative dm-meeting time. Sorour will find the best time that works for all. Thank you. Regards, Yao ================== ***Title*** Optimizing and Understanding Network Structure for Diusion *** Link of Yao's proposal*** http://people.cs.vt.edu/~yaozhang/prelim/thesis.pdf *** Abstract *** Given a population contact network, how to distribute vaccines to individuals to effectively control a flu epidemic? Similarly, in social media websites, how to choose the best communities/groups to stop rumors from spreading? Additionally, how to nd the best accounts that bridge celebrities and ordinary users? These questions are related to diffusion (aka propagation) phenomena. Diffusion can be treated as a behavior of spreading contagions (like viruses, ideas, memes, etc) on some underlying network. It is omnipresent in areas such as social media, public health, and cyber-security. Examples include diseases like flu spreading on person-to-person contact networks, memes disseminating by online adoption over online friendship networks, and malware propagating among computer networks. When a contagion spreads, network structure (like nodes/edges/groups, etc.) plays an important role in determining the outcome. For instance, a rumor, if propagated by celebrities, can possibly go viral. Similarly, an epidemic can die out quickly, if vulnerable demographic groups are successfully targeted for vaccination. Hence in this thesis, we aim to optimize and understand network structure better in light of diffusion. We optimize graph topologies by removing nodes/edges for controlling rumors/viruses from Hence in this thesis, we aim to optimize and understand network structure better in light of diffusion. We optimize graph topologies by removing nodes/edges for controlling rumors/viruses from spreading, and gain a deeper understanding of a network in terms of diffusion by exploring how nodes group together for similar roles of dissemination. We develop several novel graph mining algorithms, with different levels of granularity (node/edge level to group/community level), from model-driven and data-driven perspectives, focusing on topics like immunization on networks, graph summarization, and community detection. 1. Model-driven. Diffusion processes are usually described using mathematical models (e.g., the Independent Cascade (IC) model in social media, and the Susceptible-InfectiousRecovered (SIR) model in epidemiology). Given such models, we propose to optimize network structure for controlling propagation (the immunization problem) in several practical settings. We develop efficient algorithms for different interventions, such as vaccination (node removal) and quarantining (edge removal). In addition, we study the graph coarsening problem to obtain a better understanding of relations of nodes when a contagion is propagating. Our goal is to get a much smaller representation of a large network, while preserving its diffusive properties. 2. Data-driven. In contrast to model-based methods, data-driven approaches directly work on propagation data by relaxing modeling assumptions of diffusion. Hence they are more realistic for contagion related problems. This is due to the fact that typically diffusion processes are very complicated in reality, and it is challenging or even impossible to pick the most suited model to describe them. In this part of the thesis, we use data directly from sources like tweets and electronic healthcare records. We first propose a novel community detection problem to discover bridge" and celebrity" communities from social media data, and design case studies to understand roles of nodes/communities using diffusion. Second, we plan to develop data-driven immunization strategies to stop rumors or allocate vaccines by optimizing network topologies, using over 10 million tweets or national-level diagnostic patient data. Our work has many applications in multiple areas such as epidemiology, sociology and computer science. For example, it can help Our work has many applications in multiple areas such as epidemiology, sociology and computer science. For example, it can help discovery communities of users in terms of propagations in social media. Similarly, our work on efficient immunization algorithms can guide better policies to control epidemics.

2 2