BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:ENBIS-QSR webinar: Enhancing Robustness in Reinforcement Learning:
  Domain Knowledge Integration and Lower-Bound Certification
DTSTART:20260402T140000Z
DTEND:20260402T150000Z
DTSTAMP:20260615T023500Z
UID:indico-event-91@conferences.enbis.org
DESCRIPTION:Speakers: Bo Shen (New Jersey Institute of Technology)\n\n\nEN
 BIS-QSR webinar: Enhancing Robustness in Reinforcement Learning: Domain Kn
 owledge Integration and Lower-Bound Certification\nSpeaker: Yisha Xiang (I
 ndustrial & Systems Engineering Department at the University of Houston)\n
 Chair: Bo Shen (New Jersey Institute of Technology)\nDate: 2nd April 2026\
 , at 16:00-17:00 CEST/ 10:00-11:00 EST\nReinforcement learning (RL) has be
 come a powerful framework for solving sequential decision-making problems\
 , particularly in complex and dynamic environments where traditional optim
 ization techniques often struggle. This work develops novel RL-based metho
 dologies to address critical challenges in domain knowledge incorporation 
 and robustness certification in RL\, contributing to both theoretical adva
 ncements and practical applications. We first develop a domain knowledge-i
 nformed deep reinforcement learning method to leverage prior knowledge and
  overcome slow convergence in conventional Q-learning. Theoretical results
  guarantee convergence for small-scale problems\, while computational expe
 riments on larger problems demonstrate superior efficiency and reward perf
 ormance compared to traditional methods. The second part of this work prop
 oses a novel certification framework for evaluating the robustness of RL p
 olicies under bounded adversarial state perturbations. We formulate the lo
 wer bound certification problem under p-norm-bounded perturbations as a co
 nvex optimization problem through a phi-divergence-based relaxation\, and 
 derive its dual formulation to establish tractable\, risk-aware lower boun
 ds on the expected exponential utility of smoothed policies. We further de
 velop an empirical method to improve the certified lower bounds of RL poli
 cies. Our results show that risk-averse training generally results in poli
 cies with higher certified lower bounds than risk-neutral training\, espec
 ially under larger perturbation budgets.\n \nBio:\nDr. Yisha Xiang is an 
 Associate Professor and the Scott T. Poage Fellow in the Industrial & Syst
 ems Engineering Department at the University of Houston. Her current resea
 rch and teaching interests involve data-driven decision-making under uncer
 tainty and statistical machine learning. Her research has been funded by t
 he National Science Foundation\, including a CAREER grant\, and industry. 
 She has published articles in refereed journals\, such as INFORMS journal 
 on Computing\, IISE Transactions\, European Journal of Operational Researc
 h\, and Naval Research Logistics. She was the recipient of the P.K. McElro
 y award\, Stan Oftshun award\, and Doug Ogden award for best papers at the
  Reliability and Maintainability Symposium. Dr. Xiang received her B.S. in
  Industrial Engineering from Nanjing University of Aero. & Astro.\, China\
 , and M.S. and Ph.D. in Industrial Engineering from University of Arkansas
 . She serves as an Associate Editor for IISE Transactions\, INFORMS Journa
 l on Data Science\, and IEEE Transactions on Reliability\, and was previou
 sly on the editorial team of IEEE Transactions on Automation Science and E
 ngineering. She has served as President of the IISE Quality Control and Re
 liability Engineering Division and Chair of the INFORMS Quality\, Statisti
 cs\, and Reliability Section.  She is a senior member of INFORMS and a me
 mber of IISE.\n \n\n \n\nhttps://conferences.enbis.org/event/91/
URL:https://conferences.enbis.org/event/91/
END:VEVENT
END:VCALENDAR