Hybrid Learning Systems utilizing Sum-Product Networks

Información del proyecto

HYBSPN

Identificador del acuerdo de subvención: 797223

DOI

10.3030/797223

Proyecto cerrado

Fecha de la firma de la CE 9 Marzo 2018

Fecha de inicio 1 Marzo 2018

Fecha de finalización 31 Diciembre 2019

Financiado con arreglo a

• EXCELLENT SCIENCE - Marie Skłodowska-Curie Actions

Coste total

€ 179 166,90

Aportación de la UE

€ 179 166,90

Coordinado por

THE CHANCELLOR MASTERS AND SCHOLARS OF THE UNIVERSITY OF CAMBRIDGE
United Kingdom

Objetivo

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.

Ámbito científico (EuroSciVoc)

• ciencias naturales › informática y ciencias de la información › inteligencia artificial › inteligencia artificial generativa
• ciencias naturales › informática y ciencias de la información › inteligencia artificial › aprendizaje automático › aprendizaje profundo
• ciencias naturales › informática y ciencias de la información › inteligencia artificial › inteligencia computacional

Programa(s)

• H2020-EU.1.3. - EXCELLENT SCIENCE - Marie Skłodowska-Curie Actions PROGRAMA PRINCIPAL
  
• H2020-EU.1.3.2. - Nurturing excellence by means of cross-border and cross-sector mobility
  

Tema(s)

• MSCA-IF-2017 - Individual Fellowships
  

Régimen de financiación

MSCA-IF-EF-ST - Standard EF

Convocatoria de propuestas

H2020-MSCA-IF-2017

Coordinador

THE CHANCELLOR MASTERS AND SCHOLARS OF THE UNIVERSITY OF CAMBRIDGE

• Aportación neta de la UEn: € 179 166,90
• Dirección: TRINITY LANE THE OLD SCHOOLS ; CB2 1TN CAMBRIDGE ; Reino Unido
• Región: East of England › East Anglia › Cambridgeshire CC
• Tipo de actividad: Higher or Secondary Education Establishments
• Coste total: € 179 166,90