News

Shuo Li, Sangdon Park, Insup Lee, and Osbert Bastani received Best Paper Award from the Neural Conversational AI Workshop at the Fortieth International Conference on Machine Learning (ICML 2023).

In their paper entitled “TRAC: Trustworthy Retrieval Augmented Chatbot”, they propose a novel strategy for retrieval augmented question answering by combining conformal prediction and global multi-hypothesis testing.

Congratulations on your well-deserved achievement!

Our postdoc, Michele Caprio, was named the IJAR Young Researcher of the Year, for his promising work and deep contributions to the field of imprecise probabilities. This Award is given by the International Journal of Approximate Reasoning (IJAR) to researchers, typically under 30, who demonstrate excellence in research at an early stage of their scientific career.

We congratulate Michele for his achievement! 

Insup Lee, Warren D. Seider and Karen I. Winey are among the eight Penn scholars that have been named to the 2022 class of American Association for the Advancement of Science Fellows. They are among more than 500 researchers honored for their “scientifically and socially distinguished achievements.”

Congratulations, Insup!

PRECISE faculty, Linh Thi Xuan Phan, and a team of researchers have identified a critical security flaw in the networking approach used in aerospace and other safety-critical systems.

James Weimer co-founded Neuralert Technologies in 2019 as a spin-out from the University of Pennsylvania. Its mission is to transform the method of monitoring for stroke symptoms in hospitalized patients. Their "Stroke Detection Wristband" Named to TIME’s List of the Best Inventions of 2022. 

Congratulations!

A very warm congratulations to Mayur, Saswat Anand, Mary Jean Harrold, and Hongseok Yang! Their paper titled "Automated Concolic Testing of Smartphone Apps" was selected for the ESEC/FSE 2022 Test of Time Award.  This is a great recognition of the outstanding and impactful work he's doing!

ACM SIGBED established the Distinguished Leadership Awards to recognize individuals who have exemplary and substantive leadership in leading and implementing activities over the communities relevant to SIGBED at regional, national, and/or international level.

PRECISE Director, Insup Lee, was named the recipient of the ACM SIGBED Inaugural Distinguished Leadership Award this year for his leadership in promoting cross-fertilization of ACM and IEEE communities in Cyber-Physical Systems, Embedded Systems, and Real-Time Systems.

Warmest congratulations to Insup!

Radoslav Ivanov, Taylor J. Carpenter, James Weimer, Rajeev Alur, George J. Pappas, and Insup Lee won ACM Transactions on Embedded Computing Systems (TECS) Best Paper Award at ESWeek 2022. Their paper addresses the problem of verifying the safety of autonomous systems with neural network (NN) controllers; and provides evaluations on four benchmarks.

Congratulations on your well-deserved achievement!

PRECISE Faculty, Linh Thi Xuan Phan, will be collaborating with an international team of researchers that will work to unify the design of software and hardware components in cyber-physical transportation systems.

This five-year, $5.7 million project, is being funded by the National Science Foundation (NSF), and will be led by researchers at the University of California, Santa Cruz. It will feature collaborators from the University of California, Berkeley; Vanderbilt University; the University of Colorado, Boulder; the Norwegian University of Science Technology and Italy’s IMT School for Advanced Studies, Lucca.

A major goal of this research is to help self-driving cars and other autonomous transportation systems achieve increased efficiency and performance, as well as better safety guarantees.

Yahan Yang, Ramneet Kaur, Souradeep Dutta, and Insup Lee won the Best Paper Award at the 13th ACM/IEEE International Conference on Cyber-Physical Systems (ICCPS 2022). Below is the title and abstract for their paper.

"Interpretable Detection of Distribution Shifts in Learning Enabled Cyber-Physical Systems"

Deep neural networks have allowed us to use high dimensional real world signals generated from sensors like camera and LiDAR. However, this  comes with its potential perils. The pitfalls arise from possible over-fitting, and subsequent unsafe behavior when exposed to unknown environments. In this paper, our proposal is to build good representations for in-distribution data. We introduce the idea of a memory bank to store prototypical samples from the input space. We use these memories to compute     probability density estimates using kernel density estimation techniques. We evaluate our technique on two challenging scenarios :  a self-driving car setting implemented inside the simulator CARLA with image inputs, and an autonomous racing car navigation setting, with LiDAR inputs. An added benefit of using training samples as memories to detect out-of-distribution inputs is that the system is interpretable to a human operator. Explanation of this nature is generally hard to obtain from pure deep learning based alternatives.  

Congratulations on your well-deserved achievement!