Integrated Calendar

I will present three deep learning algorithms for registering 3D point clouds in different settings.
The first is designed to find a rigid transformation between point clouds and is based on
the concept of best buddies similarity. The second algorithm offers a fast method for non-rigid
dense correspondence between point clouds based on structured shape construction.
Finally, I extend the second algorithm to handle scene flow estimation that can be learned on a small amount
of data without employing ground-truth flow supervision.

Rhodopsins are photoreceptive membrane proteins containing a retinal chromophore in animals and microbes. Animal and microbial rhodopsins possess 11-cis and all-trans retinal, respectively, and undergo isomerization into all-trans and 13-cis retinal by light. While animal rhodopsins are G protein coupled receptors, the function of microbial rhodopsins is highly divergent, including light-driven ion pumps, light-gated ion channels, photosensors, and light-activated enzymes. Microbial rhodopsins have been the main tools in optogenetics.
Function of rhodopsins starts in 10-15 sec, and activation of rhodopsins occurs in the protein environment that has been optimized during evolution (1015 sec). We thus need various methods to understand these events of 30 orders of magnitude in time. We have studied molecular mechanism of rhodopsins by use of spectroscopic methods. Using ultrafast spectroscopy, we showed the primary event in our vision being retinal photoisomerization. In rhodopsins, photoisomerization of retinal, the shape-changing reaction, occurs even at 77 K. Using low-temperature infrared spectroscopy, we detected protein-bound water molecules of rhodopsins before X-ray crystallography. Detailed vibrational analysis provided structural information such as our color discrimination mechanism.
I will talk about our spectroscopic study of animal and microbial rhodopsins. Recent unexpected findings such as unusual isomerization pathways and temperature effects are also presented.

Since 1970, the University of Rochester and other laboratories around the world have
built more energetic and more powerful lasers. These technology advances have enabled new
science regimes. Fifty years later, fusion ignition has been achieved in the laboratory ,
where more energy than the laser energy was released ; an amazing demonstration of precision
science under extreme conditions.Astrophysics is now a laboratory science-new equations of
state, constitutive properties and structures are measured at conditions equivalent to giant gas
planets and super earths. Ultrashort pulse lasers, a LLE invention acknowledged in the 2018
Nobel Prize for Physics, is enabling ultrahigh field physics and new generations of particle
accelerators and light sources. Recent progress on ignition, high-energy-density science and
short-pulse laser physics will be summarized. The pursuit of direct-drive fusion and the path
to 25 Petawatt lasers will be discussed.

The typical approach towards drawing causal conclusions from observed data starts by defining a causal estimand, for example in terms of potential outcomes or the so-called do operator, and continues by providing conditions for identification of this estimand from the data, followed by statistical estimation and inference. One of the main assumptions is the no-interference assumption, meaning that the treatment assigned to one unit does not affect other units in the sample. However, in many domains such as in the social sciences and infectious disease epidemiology, this assumption is implausible in practice due to social interactions.
As an alternative to the no-interference assumption, an interference structure is often represented using a network. Ubiquitously, the network structure is assumed to be known and correctly specified. Nevertheless, correctly encoding the interference structure in a network can be challenging. For example, people may misreport their social connections, or report connections irrelevant to the specific combination of treatment and outcome.
Building on the exposure mapping framework, we derive the bias arising from estimating causal effects under a misspecified interference structure. To address this problem, we propose a novel estimator that uses multiple networks simultaneously and is unbiased if one of the networks correctly represents the interference structure, thus providing robustness to the network specification. Additionally, we propose a sensitivity analysis that quantifies the impact of a postulated misspecification mechanism on the causal estimates. Through simulation studies, we illustrate the bias from assuming an incorrect network and show the bias-variance tradeoff of our proposed network-misspecification-robust estimator. We further demonstrate the utility of our methods in two real examples.
Joint work with Bar Weinstein