ISI Graduate Student Symposium best poster winner Abdul Qadeer, left (not pictured: supervisor John Heidemann) and best presentation winner Sivaramakrishnan Satyamangalam Ramanathan with supervisor Jelena Mirkovic (not pictured: Minlan Yu). Photo/Caitlin Dawson.
On April 5, some of ISI’s finest research and creative projects were showcased at the Graduate Student Symposium, an annual event that celebrates the research efforts and accomplishments of students and faculty across the institute’s divisions.
The day-long event, organized by ISI graduate students, welcomed more than 40 guests at ISI’s Marina del Rey headquarters. In total, 22 students shared their research projects in oral and poster presentations on a broad variety of topics ranging from mathematical modelling to cybersecurity and artificial intelligence.
Established in 2006 by ISI’s Intelligent Systems Division, the symposium was expanded to include the Internet and Networked Systems Division in 2012, and became an institute-wide event in 2013. This year’s symposium was organized by ISI graduate students Minh Pham, Geoffrey Tran and Liang Zhu, with advisors Christopher Hauser, a research computer scientist, and Yigal Arens, ISI’s senior director of administrative affairs. Awards for best poster and best presentation were determined by audience members comprising faculty, staff and students.
“This inspiring event encourage the crucial communication skills required for successful careers in technology and fosters avenues of discussion between research groups across the institute,” said Hauser.
“We were very impressed by the caliber of student research presented at this year’s symposium and wish to congratulate the winners and everyone who took part.”
The winners:
Best poster: Abdul Qadeer, John Heidemann. Plumb: Efficient Multi-user Large-file Stream Processing in a Lumpy Pipeline of Serial Programs.
Quadeer explains: “Cost-effective and timely processing of high-value data of network data is important. Network operators often consume bulk streaming data, with multiple users consuming the data to generate statistics, detect attacks and archive data. Our Plumb processing system optimizes multi-user computation by removing duplicate processing and data storage. The Plumb execution environment improves system-wide throughput and reduces resource consumption, while providing a simple interface to end-users.”
Best presentation: Sivaramakrishnan Satyamangalam Ramanathan, Jelena Mirkovic, Minlan Yu. Blacklists Assemble: Aggregating Blacklists for Accuracy and Speed.
Ramanathan explains: “Compromised devices are constantly drafted into botnets and misused for attacks, such as sending spam and phishing emails, participating in distributed denial-of-service (DDoS) attacks and spreading malware. Filtering traffic from blacklisted sources can help prevent attacks and reduce the load on defenses such as spam filters, network intrusion detection systems and DDoS defenses. BLAG (for Blacklist Aggregator) is a sophisticated approach that selects and aggregates only accurate pieces of information from multiple blacklists. BLAG calculates information about the accuracy of each blacklist and uses recommender systems to select the most reputable and accurate pieces of information to aggregate into a master blacklist.”
Click here for a full list of presentations.
Published on April 19th, 2018
Last updated on May 20th, 2021
