Peng, Bareinboim, and Blocki Receive NSF Awards - Department of Computer Science - Purdue University Skip to main content

Peng, Bareinboim, and Blocki Receive NSF Awards

04-05-2018

Writer(s): Kristyn Childres

TrioPhoto

Chunyi Peng, Elias Bareinboim and Jeremiah Blocki, all assistant professors in the Department of Computer Science, have received awards from the National Science Foundation (NSF). Peng and Bareinboim received CAREER awards, and Blocki received a CRII award. Ben Delaware received a CRII award earlier this semester.

The CAREER program was created to support junior faculty in light of their outstanding research, excellent education and the integration of education and research within the context of the mission of their organizations. The CRII program was established to encourage research independence by supporting faculty during their first two years in an academic position.

Chunyi Peng: Amplifying Intelligence in Mobile Networked Systems
The mobile Internet revolution is ongoing. As the technology evolves from the fourth generation (4G) to 5G and beyond, there is a need to incorporate in these networks a substantially higher degree of intelligence than is present today.

“Current 4G systems don’t provide sufficient information on network behaviors, making it challenging to optimize performance and diagnose failures, and furthermore discouraging the deployment of new applications like virtual and augmented realities, and autonomous vehicles,” said Peng. Her project seeks to enhance intelligence in the 4G/5G signaling subsystem, offering network control utilities like radio resource control and mobility management.

“This work will lead to improvements in performance and reliability for our mobile Internet infrastructure, seeking to influence the upcoming 5G and post-5G standardization,” she said.

Peng will work closely with mobile network companies during this project. She will develop Amplifying Intelligence in Mobile Network Systems (AIM), a multi-disciplinary solution that applies machine learning, data science, distributed systems and computing theory to mobile networking. It offers protocol and function analytics, control-plane signaling for data access, enhanced error-handling design and mobile VR for data-plane signaling intelligence. AIM will tackle the problem of high complexity and lack of verification in the infrastructure, as well as addressing the current lack of information about underlying network operations at the level of the mobile operating system.

Elias Barenboim: Approximate Causal Inference
More data is available than ever before, giving researchers many opportunities for interesting study. In many practical, large-scale settings, it is still difficult to prove causality or determine meaningful explanations for phenomena.

In particular, the knowledge available to the scientist does not always match the formal requirements for proving causality. “These requirements cannot be waived, yet, in practice, causal claims are being made even when these conditions are not met,” said Bareinboim. He notes that there is a positive realization, too. “Throughout the empirical disciplines, there is increasing recognition that many of the scientific findings articulated today are too fragile, incapable of surviving more rigorous scrutiny or even being reproduced.”

This project will bridge the gap between the knowledge available to the researcher in practical settings and the conditions required by the theory, which, if followed, would generate robust and scientifically-grounded causal claims. The project will offer foundational grounding for most of the data science inferences made today, which will impact the practice of several data-intensive fields built on cause-and-effect relationships, including econometrics, education, bioinformatics, and medicine.

Bareinboim is also concerned with the next generation of data science researchers. The project contains an educational component targeting this audience. He proposes a new educational platform tailored to teaching causal inference concepts, principles, and tools to STEM students. The primary goal of this new platform is to move towards a more fundamental understanding of the conditions that legitimize causal statements, which should improve data literacy and lead to a more critical view of the published literature and interpretation of complex phenomena.

Jeremiah Blocki: Stronger Memory-Hard Functions for Secure Password Hashing

Recent data breaches have exposed over a billion user passwords to offline attacks. Password hash functions are the last defense against an offline attacker who has stolen password hash values from an authentication server. That attacker can attempt to crack each user’s password offline by comparing the hashes of likely password guesses with the stolen hash value. Because the attacker can check each guess offline, it is no longer possible to lock out the adversary after several incorrect guesses. The attacker is limited only by the cost of computing the hash function.

Currently deployed password hashing algorithms do not provide adequate protection against these attacks. However, memory hard functions (MHFs), functions whose computation requires a large amount of memory, are a crucial cryptographic tool to achieve this goal. A main goal of this work is to establish a scientific basis for tuning the parameters of an MHF to ensure that brute-force attacks are prohibitively expensive for attackers.

Blocki will develop improved cryptanalysis techniques to evaluate MHF security, resulting in a deeper understanding of currently deployed MHFs. This project will also directly impact future cryptographic standards for password hashing by creating more secure standards through MHFs.

Bareinboim will develop the necessary identification conditions to accept as input a model that is not fully specified (where only a coarse description of the model is available). He will develop effective procedures for determining whether causality can be approximated. The project will leverage results to design efficient learning algorithms under the relaxed assumption that an input is only partially specified and an output can be an approximation of the target causal distribution.

Last Updated: May 4, 2018 2:21 PM

Department of Computer Science, 305 N. University Street, West Lafayette, IN 47907

Purdue University Indianapolis, 723 W. Michigan St., Indianapolis, IN 46202

Phone: (765) 494-6010 • Fax: (765) 494-0739

Copyright © 2024 Purdue University | An equal access/equal opportunity university | Copyright Complaints | DOE Degree Scorecards

Trouble with this page? Accessibility issues? Please contact the College of Science.