Event Date
SPEAKER: Xiucai Ding, Assistant Professor, Statistics, UC Davis
TITLE: "Phantom hunting for machine learning: high dimensional statistical analysis of data fusion algorithms"
ABSTRACT: In the past decades, we have witnessed the success of machine learning and deep learning in advancing science and engineering. However, most of the algorithms are lack of theoretical foundation so that the interpretation of the results is not guaranteed to be correct. In this talk, we study two famous kernel- based data fusion algorithms. We show that if these algorithms are applied directly without any sanity check, some phantoms (aka artifacts) can be hunted. Specifically, we prove that even though the data equipment only collects noise, the outputs of these algorithms can be misunderstood as signals; or even though the inputs have strong signals, the algorithms can be blind to these signals and report nothing. For the latter case, we provide an adaptive method to choose the bandwidth to avoid these issues. This rings the bell for the users that a careful checking and inference needed to be done before applying these algos. Similar issues for other famous algos like LLE and t-SNE will also be discussed.
DATE: Thursday October 21st, 2021
LOCATION:
MSB 1147, Colloquium Room*
*This will be an in-person seminar and is open to the public. STA 290 registrants are required to attend in person. For others, there will be a Zoom link available if you choose to listen in to the seminar remotely. To access the Zoom meeting for this seminar, please contact the instructor Professor Drake or Pete Scully ([email protected]) for the meeting ID and password, stating your affiliation.