Leading the way for the awards was a paper titled, “Firebird: Predicting Fire Risk and Prioritizing Fire Inspections in Atlanta,” in the Applied Data Science track. Stemming from collaborative research done in 2015 by Data Science for Social Good (DSSG) Atlanta fellows and the Atlanta Fire Rescue Department, the paper was selected as the “Best Student Paper Award” runner-up.

Categorizing fire risk with data analysis

The paper detailed how the team, mentored by School of Computational Science and Engineering Assistant Professors Bistra Dilkina and Polo Chau, leveraged machine learning, geocoding, and information visualization, to develop a framework to help fire departments identify buildings and businesses that need to be inspected for fire safety.

The team also developed a methodology for prioritizing commercial properties based on fire risk. Firebird computed fire risk scores for more than 5,000 buildings in the Atlanta area, and it identified 6,096 new commercial properties that potentially need to be inspected for fire safety.

Firebird has already made a positive impact on both local and national levels. It has improved AFRD’s inspection process, allowing for more informed decisions and enhanced fire safety for Atlanta’s residents. Additionally, the program was highlighted by the National Fire Protection Association (NFPA) as a best practice for using data to for fire inspections.

“Effective fire inspection targeting is a genuine problem that plagues many fire departments,” said Dilkina. “We think Firebird is a potent tool that more cities should adopt to help prioritize their fire inspection burden while protecting economic interests and lives from costly fire destruction.”

Along with the Firebird research paper award, CSE postdoctoral researcher Mohammad Taha Bahadori was named a 2016 SIGKDD Dissertation Award runner up. Bahadori earned the recognition for his work with scalable multivariate time series analysis.

Six other Georgia Tech papers, which are listed below, were presented at KDD 2016.

#GTComputing contingent goes big

In all, more than 20 College of Computing students, faculty, and staff attended this year’s conference. Highlights of the event were presentations from Assistant Professor Le Song and Professor Irfan Essa from the School of Computational Science and Engineering (CSE) and the School of Interactive Computing, respectively.

Song’s address, titled "Dynamic Processes over Information Networks Representation, Modeling, Learning and Inference," opened the Workshop on Mining and Learning From Time Series. During the Workshop on Large Scale Sports Analytics, Essa presented findings on the relationship between sports and video when analyzing athletic performances.

Along with the keynote addresses, and paper and poster presentations, the Georgia Tech team also staffed an exhibit booth during KDD. The booth highlighted Tech's Institute for Data Engineering and Science, the OMS CS program, MS Analytics program, and the new Center for Machine Learning.

In addition, Chau co-organized a full-day Interactive Data Exploration and Analytics (IDEA) workshop, sponsored by Microsoft Research and Trifacta. The workshop featured keynote speakers from Stanford, Microsoft Research, University of Washington, and the University of Calif. - Riverside.

KDD is a premier interdisciplinary conference, brings together researchers and practitioners from data science, data mining, knowledge discovery, large-scale data analytics, and big data.

Georgia Tech research papers at KDD 2016

Lexis: An Optimization Framework for Discovering the Hierarchical Structure of Sequential Data

Author(s): Payam Siyari*, Georgia Institute of Technology; Bistra Dilkina, Georgia Tech; Constantine Dovrolis, Georgia Institute of Technology

Recurrent Marked Temporal Point Processes: Embedding Event History to Vector

Author(s): Nan Du, Georgia Institute of Technology; Hanjun Dai, Georgia Institute of Technology; Rakshit Trivedi, Georgia Institute of Technology; Utkarsh Upadhyay, Max Plank Institute; Manuel Gomez-Rodriguez, MPI-SWS; Le Song, Georgia Institute of Technology 

FLASH: Fast Bayesian Optimization for Data Analytic Pipelines

Author(s): Yuyu Zhang*, Georgia Institute of Technology; Mohammad Bahadori, Georgia Institute of Technology; Hang Su, Georgia Institute of Technology; Jimeng Sun, Georgia Institute of Technology 

Multi-layer Representation Learning for Medical Concepts

Author(s): Edward Choi*, Georgia Institute of Technology; Mohammad Bahadori, Georgia Institute of Technology; Elizabeth Searles, Children's Healthcare of Atlanta; Catherine Coffey, Children's Healthcare of Atlanta; Michael Thompson, Children's Healthcare of Atlanta; James Bost, Children's Healthcare of Atlanta; Javier Tejedor-Sojo, Children's Healthcare of Atlanta; Jimeng Sun, Georgia Institute of Technology

Communication Efficient Distributed Kernel Principal Component Analysis

Author(s): Yingyu Liang*, Princeton University; Bo Xie, Georgia Institute of Technology; David Woodruff, IBM Research; Le Song, Georgia Institute of Technology; Maria-Florina Balcan, Carnegie Mellon University

Smart broadcasting: Do you want to be seen?

Author(s): Erfan Tavakoli, Sharif University; Mohammad Reza Karimi, Sharif University; Mehrdad Farajtabar, Georgia Institute of Technology; Le Song, Georgia Institute of Technology; Manuel Gomez-Rodriguez*, MPI-SWS