Opendata, web and dolomites

HYBSPN SIGNED

Hybrid Learning Systems utilizing Sum-Product Networks

Total Cost €

0

EC-Contrib. €

0

Partnership

0

Views

0

Project "HYBSPN" data sheet

The following table provides information about the project.

Coordinator
THE CHANCELLOR MASTERS AND SCHOLARSOF THE UNIVERSITY OF CAMBRIDGE 

Organization address
address: TRINITY LANE THE OLD SCHOOLS
city: CAMBRIDGE
postcode: CB2 1TN
website: www.cam.ac.uk

contact info
title: n.a.
name: n.a.
surname: n.a.
function: n.a.
email: n.a.
telephone: n.a.
fax: n.a.

 Coordinator Country United Kingdom [UK]
 Total cost 179˙166 €
 EC max contribution 179˙166 € (100%)
 Programme 1. H2020-EU.1.3.2. (Nurturing excellence by means of cross-border and cross-sector mobility)
 Code Call H2020-MSCA-IF-2017
 Funding Scheme MSCA-IF-EF-ST
 Starting year 2018
 Duration (year-month-day) from 2018-03-01   to  2019-12-31

 Partnership

Take a look of project's partnership.

# participants  country  role  EC contrib. [€] 
1    THE CHANCELLOR MASTERS AND SCHOLARSOF THE UNIVERSITY OF CAMBRIDGE UK (CAMBRIDGE) coordinator 179˙166.00

Map

 Project objective

We have recently witnessed a considerable interest in probabilistic models within deep learning, leading to e.g. generative adversarial networks, deep generative networks, neural auto-regressive density estimators and Pixel-RNNs/CNNs. Furthermore, sum-product networks (SPNs) are a recent deep architecture with a unique advantage over the aforementioned models: they allow both exact and efficient inference, implemented in terms of simple network passes. However, SPNs are a constrained type of neural network and do not reach the full flexibility of the deep learning tool kit available to date. This calls for hybrid learning systems which exploit the superior inference properties of SPNs within other deep learning approaches. In this project, I will investigate two such approaches. First, I will structurally combine a deep learning architecture (front-end), which extracts a representation from a set of inputs, controlling the parameters of an SPN (back-end) over a set of outputs. This yields a hybrid conditional SPN which facilitates full inference over the output space, and which is naturally applied in structural prediction tasks. Such hybrid SPNs can be expected to be highly expressive and to set new state-of-the-art results in e.g. semantic image segmentation. The second approach is to use SPNs as variational distributions, i.e. for approximating a given target distribution by minimizing Kullback-Leibler divergence. On the one hand, this allows to capture intractable models with SPNs, with the goal to enable fast amortized approximate inference. On the other hand, this approach allows to use hybrid conditional SPNs as so-called inference networks for intractable generative models with latent variables, for the purpose of variational posterior inference and learning. This approach would represent a substantial improvement over state-of-the-art approaches, which are usually limited to expensive inference via Monte Carlo estimation.

 Publications

year authors and title journal last update
List of publications.
2019 Martin Trapp, Robert Peharz, Hong Ge, Franz Pernkopf, Zoubin Ghahramani
Bayesian Learning of Sum-Product Networks
published pages: 6344--6355, ISSN: , DOI:
Advances in Neural Information Processing Systems 32 2020-03-06
2018 Molina, Alejandro; Vergari, Antonio; Stelzner, Karl; Peharz, Robert; Subramani, Pranav; Di Mauro, Nicola; Poupart, Pascal; Kersting, Kristian
SPFlow: An Easy and Extensible Library for Deep Probabilistic Learning using Sum-Product Networks
published pages: , ISSN: , DOI:
Arxiv preprints 1 2020-03-06
2019 Martin Trapp, Robert Peharz, Franz Pernkopf, Carl E. Rasmussen
Deep Structured Mixtures of Gaussian Processes
published pages: , ISSN: , DOI:
Arxiv preprints 2020-03-06
2018 Wolfgang Roth, Robert Peharz, Sebastian Tschiatschek, Franz Pernkopf
Hybrid generative-discriminative training of Gaussian mixture models
published pages: 131-137, ISSN: 0167-8655, DOI: 10.1016/j.patrec.2018.06.014
Pattern Recognition Letters 112 2020-03-06
2018 Pernkopf, Franz; Roth, Wolfgang; Zoehrer, Matthias; Pfeifenberger, Lukas; Schindler, Guenther; Froening, Holger; Tschiatschek, Sebastian; Peharz, Robert; Mattina, Matthew; Ghahramani, Zoubin
Efficient and Robust Machine Learning for Real-World Systems
published pages: , ISSN: , DOI:
Arxiv preprints 1 2020-03-06
2019 Xiaoting Shao, Alejandro Molina, Antonio Vergari, Karl Stelzner, Robert Peharz, Thomas Liebig, Kristian Kersting
Conditional Sum-Product Networks: Imposing Structure on Deep Probabilistic Architectures
published pages: , ISSN: , DOI:
Arxiv Preprints 2020-03-06
2019 Cory J. Butz, Jhonatan S. Oliveira, Robert Peharz
Sum-Product Network Decompilation
published pages: , ISSN: , DOI:
Arxiv preprints 2020-01-29

Are you the coordinator (or a participant) of this project? Plaese send me more information about the "HYBSPN" project.

For instance: the website url (it has not provided by EU-opendata yet), the logo, a more detailed description of the project (in plain text as a rtf file or a word file), some pictures (as picture files, not embedded into any word file), twitter account, linkedin page, etc.

Send me an  email (fabio@fabiodisconzi.com) and I put them in your project's page as son as possible.

Thanks. And then put a link of this page into your project's website.

The information about "HYBSPN" are provided by the European Opendata Portal: CORDIS opendata.

More projects from the same programme (H2020-EU.1.3.2.)

Self-EsteemProcesses (2020)

A self-esteem process framework of the transition to work

Read More  

STIMOS (2019)

Stimulation of Multiple Organoids Simultaneously

Read More  

INTERGLP1 (2020)

Dissecting GLP-1 receptor internalization pathways using genetic and pharmacological tools

Read More