Millennium Café

Physics-based models and purely data-driven machine learning models each have significant benefits and limitations.  A new pathway, differentiable modelling, combines the two methods and pushes the boundary of physics-informed machine learning.  I will use some examples from water resource research to demonstrate advantages of this combined approach where we have mitigated the limitations of each method, more reliably predicted geohazards like floods, and discovered previously unrecognized physical relationships.

Presenter: Chaopeng Shen  |  Civil & Environmental Engineering

The cold universe and the sun are two important renewable energy resources. We have developed a device that can passively cool to about 5 ˚C below the ambient temperature via radiative cooling and generate electricity using sunlight at the same time and same place. I’ll use this example to highlight novel opportunities for harvesting renewable energy sources.

Linxiao Zhu  |  Mechanical Engineering

ChatGPT can craft essays that appear human-made, significantly influencing how liberal arts and humanities are taught and graded. Soon, its impacts will be felt by the world of science and engineering. In this talk, I'll showcase how ChatGPT can produce convincing and thorough reviews of scientific papers and proposals, using my past IEE seed grants as case studies. These AI-generated reviews sound authentic, but it's uncertain if they match the quantitative rankings human experts would provide. Moreover, as ChatGPT reshapes the review process, we face a new challenge: discerning whether reviewers are genuinely offering their expertise or leaning on computer tools. The title and abstract for this talk were generated by ChatGPT 4. 

Speaker: Chris Gorski  |  Civil & Environmental Engineering

The International Atomic Energy Agency defines nuclear forensics as …the examination of nuclear or other radioactive material, or of evidence that is contaminated (or comingled) with radionuclides, in the context of legal proceedings under international or national laws related to nuclear security. Historically, isotopic analyses of uranium and plutonium have been relied upon to provide key insights into the reactor or enrichment operations used to create special nuclear materials found outside of regulatory control (MORC). However, isotopic information provides no clues to the chemical process history of unknown materials. For this, scientists are attempting to exploit, among other things, morphological and microstructural features to reconstruct the process history of MORC…but many challenges remain.

Speaker: Jon Schwantes  |  Nuclear Engineering   

Penn State and Monash University (in Melbourne, Australia) share some important characteristics: our intellectual culture and our research strengths are very well aligned. Penn State and Monash are now official partners, with a Memorandum of Understanding having been signed last September by President Neeli Bendapudi and Monash President Margaret Gardner (a good question to ask will be “where is Margaret now?”). This partnership opens up a world of opportunities for researchers at both universities. I've just returned from a visit to Monash and will share with you what I've learned and as much as I can from your questions. Get ready to pack your bags and to taste test Vegemite!

Speaker: Jenni Evans  |  Director, Institute for Computational and Data Sciences

I will overview recent efforts towards developing methods to assess confidence in published findings in the literature using AI.  These efforts serve the broader aim of supporting efficient scientific processes in light of the last decade’s replication crisis and ongoing open science movement. We will end with the vision of a computable scholarly record.

Speaker: Sarah Rajtmajer  |  College of Information Sciences & Technology