Search Constraints
Number of results to display per page
Results for:
Affiliation
Australian National University
Remove constraint Affiliation: Australian National University
1 - 2 of 2
Search Results
-
MOOC
Interested in exploring the deadliest and most mysterious parts of our universe? Or, investigating black holes, which warp the very fabric of space-time around them? We will look at what we know about these objects, and also at the many unsolved mysteries that surround them. We will also study white-dwarf stars and neutron stars, where the mind-bending laws of quantum mechanics collide with relativity. And, examine dwarf novae, classical novae, supernovae and even hypernovae: the most violent explosions in the cosmos. This course is designed for people who would like to get a deeper understanding of astronomy than that offered by popular science articles and television shows.You will need reasonable high-school level Maths and Physics to get the most out of this course.
- Course related:
- AP1D02 Introduction to Astronomy
- Subjects:
- Environmental Sciences
- Keywords:
- Black holes (Astronomy) Astrophysics Collisions (Astrophysics)
- Resource Type:
- MOOC
-
Presentation
This video was recorded at European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD), Athens 2011. Hierarchical modeling and reasoning are fundamental in machine intelligence, and for this the two-parameter Poisson-Dirichlet Process (PDP) plays an important role. The most popular MCMC sampling algorithm for the hierarchical PDP and hierarchical Dirichlet Process is to conduct an incremental sampling based on the Chinese restaurant metaphor, which originates from the Chinese restaurant process (CRP). In this paper, with the same metaphor, we propose a new table representation for the hierarchical PDPs by introducing an auxiliary latent variable, called table indicator, to record which customer takes responsibility for starting a new table. In this way, the new representation allows full exchangeability that is an essential condition for a correct Gibbs sampling algorithm. Based on this representation, we develop a block Gibbs sampling algorithm, which can jointly sample the data item and its table contribution. We test this out on the hierarchical Dirichlet process variant of latent Dirichlet allocation (HDP-LDA) developed by Teh, Jordan, Beal and Blei. Experiment results show that the proposed algorithm outperforms their "posterior sampling by direct assignment" algorithm in both out-of-sample perplexity and convergence speed. The representation can be used with many other hierarchical PDP models.
- Subjects:
- Computing
- Keywords:
- Machine learning Artificial intelligence
- Resource Type:
- Presentation