• A community day for families and people of all ages filled with balloon racing, water tables, and wind sculptures. Open to all of the Tucson community, ages 2-100.

The standard machine learning methods, called "supervised learning", take in a dataset passively and then build a model that can make accurate predictions for future data.  In many situations, however, we can choose actively which data to collect (or desire to do so to maximally use the budget).  That is, we may collect data wisely (e.g., adaptive experiments) so we use significantly less data while achieving the same performance (e.g., identification of interesting genes).  At the same time, adaptive data collection means that we are breaking the standard i.i.d. assumption on the data, which is a significant challenge as theorems and principles developed for supervised learning do not apply here.  In this talk, I will talk about novel adaptive data collection and learning algorithms arising from the so-called multi-armed bandit framework and show their theoretical guarantees and their effectiveness in real-world applications. Specifically, I will first show how biological experiments can be performed with a significantly less budget by adaptively selecting what experiments to run next.  And then, I will talk about how to accelerate drug discovery rate with bilinear models where we use drug and protein information to guide the search.

Throughout the talk, I will show various other practical  applications where bandit methods have proved valuable including AlphGo, hyperparameter tuning, and cartoon caption contest.

The Picture Rosters Tool Suite provides picture flash cards to assist instructors with learning to recognize students in the class. This tool also provides a random student selector to help ensure that students are called upon evenly in class or discussions. 

The early reporting system through D2L allows instructors to provide feedback to students enrolled in their courses. Instructors can provide both positive and constructive feedback on course performance and attendance. Students receive an email that provides them the chance to acknowledge they received your feedback. 

At this "Tools You Can Use" event, we will give a demo of the features in the Picture Roster Tool Suite as well as how to use the progress report button in D2L. Bring a laptop if you would like to try it for yourself!

Several novel hybrid imaging modalities rely on propagating sound waves to deliver spatial information about the properties of the object. Traditional image reconstruction techniques assume that an infinitely wide range of acoustic wavelengths is available. However, commonly utilized piezoelectric transducers emit/receive only a narrow band of frequencies, with both very long and very short wavelengths missing. This leads to a strong distortion in the reconstructed images. We will investigate the use of sparsity-enforcing techniques (such as, for example, L1-regularization) to overcome this problem. (Joint work with N.Do and P.Hoskins)

The Quot scheme of smooth projective surfaces parametrizing quotients of dimension 0 and 1 admit natural 2-term perfect obstruction theories. I will explain results and conjectures concerning the associated virtual invariants, with emphasis on the virtual Euler characteristics. This is based on joint work with Rahul Pandharipande.

This talk describes two experiments in traffic flow control that involve the University of Arizona CAT Vehicle Testbed. The first experiment explores how to dampen emerging waves in traffic that are due to congestive effects. This experiment grew out of theory of how traffic flow could be improved through sparse velocity control (e.g., ~5% of the vehicles) in the flow. The second experiment examines an analogous case, where 100% of the vehicles are controlled, though this time using off-the-shelf (rather than customized) cruise control algorithms. The talk will examine the hypotheses, methods, and results of these experiments, and explore the theory and motivation for the research as a means to provide insights into the obtained results. The talk will discuss how some results promise the potential for tremendous societal reduction in stress and in fuel used, and how other results indicate the need for more research to determine whether deployment at a societal scale of the available technology would result in positive, or negative, benefits in terms of safety and congestion. The research was sponsored by the National Science Foundation under award CNS-1446435, and is collaborative work with Benedetto Piccoli, Benjamin Seibold, and Dan Work.

Machine learning and artificial intelligence algorithms are now being used to automate the discovery of governing physical equations and coordinate systems from measurement data alone. However, positing a universal physical law from data is challenging:  (i) An appropriate coordinate system must also be advocated and (ii) simultaneously proposing an accompanying discrepancy model to account for the inevitable mismatch between theory and measurements must be considered.  Using a combination of deep learning and sparse regression, specifically the sparse identification of nonlinear dynamics (SINDy) algorithm, we show how a robust mathematical infrastructure can be formulated for simultaneously learning physics models and their coordinate systems.  This can be done with limited data and sensors.  We demonstrate the methods on a diverse number of examples, showing how data can maximally be exploited for scientific and engineering applications. The work also highlights the fact that the naive application of ML/AI will generally be insufficient to extract universal physical laws without further modification.

Teachers, students, and their families will join the CRR for a night of fun, playing math, logic, and board games.

Refreshments will be served.

More information at https://crr.math.arizona.edu/events-0