News Service, Kevin Brown

Perry Samson

By | | No Comments

I developed LectureTools with NSF support in response to a need to increase opportunities for student participation in larger lecture courses. It was subsequently spun off campus using NSF SBIR funding and was acquired by Echo360 which has incorporated it into its Active Learning Platform (ALP).  ALP collects data on how students behave before, during and after class including how many slides they view, how many notes they type, how many questions they answer and how many gradable questions they get correct as well as what question they pose and how often do they indicate confusion.

These unique data are used to understand how student participation is related to exam grades and to build models to forecast which students will have trouble in class far earlier in the semester.  My goal is to combine data from ALP with other data sets to ascertain which, if any, participation data allows the best prediction of student success.

Exam grades as a function of the number of words typed in students notes.

Exam grades as a function of the number of words typed in students notes.

quentinstoutfaculty13

Quentin Stout

By | | No Comments

I primarily work on developing scalable parallel algorithms to solve large scientific problems. This has been done with teams from several different disciplines and application areas. I’m most concerned with algorithms emphasizing in-memory approaches. Another area of research has developed serial algorithms for nonparametric regression. This is a flexible form of regression that only assumes a general shape, such as upward, rather than a parametric form such as linear. It can be applied to a range of learning and classification problems, such as taxonomy trees. I also work some in adaptive learning, designing efficient sampling procedures.

posselt-small

Derek Posselt

By | | No Comments

My research program utilizes several data and computing intensive tools to explore the interaction between clouds and the Earth’s climate system. These include: simulation of atmospheric processes over 1000’s of km length scales at sub-1 km horizontal grid spacing, multivariate and multisensor remote sensing measurements, and nonlinear ensemble-based data assimilation methods. Each of these requires the effective use of large-capacity computational resources. Examination of the interaction between small scale cloud processes (on the order of 10 km) and the larger scale climate system dynamics (on the order of 1000s of km) requires model domains that can simultaneously resolve planetary and cloud scale processes. Assimilation of observations for highly nonlinear processes require not just a single model integration, but large ensembles of simulations. The effective exploration of the information content of observations in systems with large numbers of degrees of freedom (e.g., retrieval of three dimensional volumes of cloud and precipitation microphysical properties), involves exploration of a multidimensional probability space. Such an exercise typically requires a very large (on the order of 10^7) number of integrations of forward algorithms that map sets of atmospheric properties to simulated satellite or ground-based radiance measurements.

Data science applications: Interaction between precipitating cloud systems and a changing climate, optimal observation of cloud properties from space, model uncertainty quantification