Probabilistic Latent Variable Decompositions for Image and Audio Analysis

Channel:
Microsoft Research
Subscribers:
344,000
Published on ● Video Link: https://www.youtube.com/watch?v=35oENZQsGH4


Duration: 1:32:05
362 views
3

In this talk we present a model which can decompose probability densities into sets of shift invariant components. We will show how this model is very well suited for audio and image problems and will demonstrate its applications on complex data sets. We will also explore the relationship of this model to well known decomposition methods (such as PCA/ICA/NMF/PARAFAC) and argue about its appropriateness when dealing with density-like representations. We will also present various extensions to this model to allow it to perform sparse coding, discover Markovian components, model a priori dispositions and temporal relationships through standard statistical models.



Other Videos By Microsoft Research


2016-09-06The Bilateral Grid and a Topological Approach to Image Segmentation
2016-09-06Software Development Practices and Knowledge Sharing: A Comparison of XP & Waterfall Team Behaviors
2016-09-06How likely is BuffonΓÇÖs needle to meet a Cantor square?
2016-09-06Dense triangle-free digraphs
2016-09-06MOP: A Generic and Efficient Runtime Verification Framework
2016-09-06An Examination of User Behaviour during Web Information Tasks
2016-09-06Modeling Science: Topic models of Scientific Journals and Other Large Document Collections
2016-09-06Exhaustive Phase Order Search Space Exploration and Evaluation
2016-09-06Supervised Dimensionality Reduction with Principal Component Analysis
2016-09-06Network Market Design for Efficient Resource Allocation
2016-09-06Probabilistic Latent Variable Decompositions for Image and Audio Analysis
2016-09-06Improving Software Security with Precise Static and Runtime Analysis
2016-09-06What Analytical Performance Modeling Teaches Us About Computer Systems Design
2016-09-06A search engine for the real world, or, a top-down approach to vision
2016-09-06Approximate Inference Techniques for Identity Uncertainty
2016-09-06Edgenet 2006 - Issues in Enterprise Networks
2016-09-06Towards a Memory Model for C++
2016-09-06Edgenet 2006 - Keynote - Model-Based Management of Distributed Services
2016-09-06SCS '06 - Wrap-up and Next Steps
2016-09-06Technology Matters
2016-09-06SCS '06 - Closing Keynotes - Part 2


Tags:
microsoft research