BEGIN:VCALENDAR
VERSION:2.0
CALSCALE:GREGORIAN
PRODID:iCalendar-Ruby
BEGIN:VEVENT
CATEGORIES:Thesis & Dissertation Defense
DESCRIPTION:Abstract\nDecision trees are considered to be one of the oldest
machine learning models which received a lot of attention from practitione
rs and research community. They are conceptually simple yet powerful. State
of-the-art frameworks\, such as XGBoost or LightGBM\, rely on them as base
learners\, but they have been used as well as standalone predictors. Despi
te the rich history of decision trees and existence of numerous methods\, t
heir applicability beyond traditional supervised learning has been explored
in limited extent. For instance\, various fast growing ML subfields\, such
as semi-supervised and selfsupervised learning\, nonlinear dimensionality
reduction\, etc. have been barely used with trees. What is common to most o
f these tasks is that the objective function takes a certain form\, which i
nvolves "manifold regularization" to exploit the geometry of the underlying
data distribution. In this dissertation\, we study decision trees and\, mo
re generally\, tree-based models under this setting. We argue that these ty
pe of problems carry a great practical importance. Furthermore\, using semi
supervised learning and nonlinear dimensionality reduction as examples\, we
derive a generic algorithm to solve such optimization problems. It is base
d on a reformulation of the problem which requires\niteratively solving two
simpler problems. One problem will always involve supervised learning of a
tree. The second one will depend on a particular problem type and can be o
ne of the following: solving a linear system\, optimizing non-linear embedd
ings or something else. We also show that the algorithm is scalable and hig
hly effective on number of tasks.\n\n \n\nBiography\n\nArman Zharmagambetov
is a PhD candidate in EECS department at UC Merced. He holds BSc and MSc i
n Mathematical and Computer Modeling from IITU (Almaty\, Kazakhstan). His p
rimary research interests are in machine learning. Specifically\, his recen
t works include developing learning algorithms for decision trees and treeb
ased models\; and their applications in various domains\, such as: (semi-)s
upervised learning\, neural net compression\, dimensionality reduction and
model interpretability. Arman is recipient of\nseveral graduate fellowships
and his publications appeared in top venues\, including NeurIPS\, ICML\, E
MNLP\, ICASSP\, etc.
DTEND:20221121T224500Z
DTSTAMP:20230323T085349Z
DTSTART:20221121T210000Z
GEO:37.363734;-120.43111
LOCATION:Science and Engineering II (S&E2)\, 302
SEQUENCE:0
SUMMARY:Electrical Engineering and Computer Science (EECS) Ph.D. Dissertati
on Defense: Arman Zharmagambetov
UID:tag:localist.com\,2008:EventInstance_41554912048157
URL:https://events.ucmerced.edu/event/electrical_engineering_and_computer_s
cience_eecs_phd_dissertation_defense_arman_zharmagambetov
END:VEVENT
END:VCALENDAR