P
Padhraic Smyth
Researcher at University of California, Irvine
Publications - 359
Citations - 38795
Padhraic Smyth is an academic researcher from University of California, Irvine. The author has contributed to research in topics: Inference & Topic model. The author has an hindex of 80, co-authored 342 publications receiving 36653 citations. Previous affiliations of Padhraic Smyth include University of California & Jet Propulsion Laboratory.
Papers
More filters
Posted Content
Hot Swapping for Online Adaptation of Optimization Hyperparameters
TL;DR: A general framework for online adaptation of optimization hyperparameters by `hot swapping' their values during learning by investigating this approach in the context of adaptive learning rate selection using an explore-exploit strategy from the multi-armed bandit literature.
Variational message-passing: extension to continuous variables and applications in multi-target tracking
TL;DR: By constructing a factor graph representation of the track-oriented multiple hypothesis tracker, this dissertation enables the application of variational inference algorithms to efficiently estimate marginal probabilities of possible tracks and shows that these track marginals are the key ingredient in a multi-target generalization of the standard expectation-maximization algorithm used for parameter estimation in single-target tracking.
Proceedings ArticleDOI
Predictive Querying for Autoregressive Neural Sequence Models
TL;DR: A general typology for predictive queries in neural autoregressive sequence models is introduced and it is shown that such queries can be systematically represented by sets of elementary building blocks and used to develop new query estimation methods based on beam search, importance sampling, and hybrids.
Fault Detection Using a Two-Model Test for Changes in the Parameters of an Autoregressive Time Series
P. Scholtz,Padhraic Smyth +1 more
TL;DR: An investigation of a statistical hypothesis testing method for detecting changes in the characteristics of an observed time series, based on a measure of the information theoretic distance between two autoregressive models.