Dr. Steven Fernandes and his research team proposed a novel confidence metric called the attribution-based confidence (ABC) metric for deep neural networks (DNNs). The ABC metric characterizes whether the output of a DNN on an input can be trusted. DNNs are known to be brittle on inputs outside their training distribution and are hence susceptible to adversarial attacks.
Dr. John J. Sunderland, PhD, MBA
Professor of Radiology-Division of Nuclear Medicine
Carver College of Medicine
University of Iowa
Abstract: The use of radioactive decay and their particulate and gamma-ray emissions in medical imaging and therapy dates back to the late 1930’s with the use of radioactive Iodine. Use of nuclear medicine expanded substantially in the 1960’s with the advent of the gamma camera, and then scientific excitement was boosted again with the invention of positron emission tomography (PET scanning) in the late 70’s. These nuclear technologies demonstrated the ability not to image the anatomy (like x-rays, CT, and later MRI), but to image the actual molecular biochemical underpinnings of diseases, like cancer (the Warburg Effect – look it up!), heart disease, and Alzheimer’s disease.
Clinical use of PET imaging began is the early 1990’s. Creighton University had one of the first clinical PET facilities in the US, opening in 1991 on Dorcas Street, complete with its own cyclotron used to produce radioactive 18F, 11C, 13N, 15O. But challenges to Medicare and insurance reimbursement coupled with regulatory complexities, mostly from FDA, resulted in slow growth, and even stagnation of the field.
Beginning around 2012, through advances in radiation detector technology, computing power, corporate investment, and infrastructure building, nuclear imaging and in particular, radiopharmaceutical therapy have taken off into one of the fastest growing segments of medicine today.
Guest Speaker: Dr. David Meyer, Professor of Mathematics at the University of California, San Diego
Quantum computing offers the possibility of efficient solutions to problems that may be classically intractable. Recently, these problems have included sampling from a probability distribution and inferring a relation from partial information (e.g., a recommender system), both of which fall into the class of data science or machine learning problems. An important characteristic of such problems and their solutions is the model for the data generating process, specifically whether it is classical or quantum. In the second part of the talk, we’ll analyze some data from a simple human behavior, answering survey questions, and discuss whether or not to model it quantum mechanically. I will (re)introduce all the quantum mechanics necessary for our analysis.
Public Lecture by Dr. Bharat Ratra, Distinguished Professor of Physics at Kansas State University.
Dark energy is the leading candidate for the mechanism that is responsible for causing the cosmological expansion to accelerate. Dr. Bharat Ratra will describe the astronomical data which persuade cosmologists that (as yet undetected) dark energy and dark matter are by far the main components of the energy budget of the universe at the present time. He will review how these observations have led to the development of a quantitative "standard" model of cosmology that describes the evolution of the universe from an early epoch of inflation to the complex hierarchy of structure seen today. In this non-technical talk, he will also discuss the basic physics, and the history of ideas, on which this model is based.
Dr. Tim Gay, Department of Physics, University of Nebraska – Lincoln
Thursday, January 29nd, 2015: 4:00 p.m., Hixon-Lied Science Building, 244
Until 1957, scientists thought that the fundamental laws of Nature must be the same whether they were applied to our Universe or the Universe that is a mirror reflection of our own. The implications of the discovery that this is not true - essentially that Nature is "handed" - will be discussed. Some interesting applications of handedness, or "chirality" in agriculture, biology, chemistry, and physics will be presented. I will also talk about some new physics experiments on chirality that may shed light on how life began on this planet.
Dr. Gay’s group is interested in polarized electron physics. Their work involves studies of polarized electrons scattering from atomic and chiral molecular targets, the development of novel sources of polarized electrons and electron polarimeters, and investigations of the fundamental nature of the electron.
Evidence from multiple indirect measurements implies that 80% of the mass in the universe is dark, non-baryonic and hence is composed of a new type of undiscovered particles. I will describe why Weakly Interacting Massive Particles (WIMPs) are the most popular candidate for the dark matter and describe how WIMPs would interact in a detector. I will then describe a leading experiment attempting to directly detect WIMP interactions, the Super Cryogenic Dark Matter Search (SuperCDMS) and present results.
Dr. Sander is interested in finding evidence of new physics. Towards that end, he is a primary investigator on the SuperCDMS collaboration looking to directly detect dark matter. He is also working on developing new detection techniques for the next generation of rare event searches.
