PORTALE DELLA DIDATTICA

Ricerca CERCA
  KEYWORD

Integrating high-level semantics and deep learning for image understanding

Reference persons FABRIZIO LAMBERTI

External reference persons Lia Morra

Research Groups GR-09 - GRAphics and INtelligent Systems - GRAINS

Thesis type RESEARCH THESIS

Description Artificial Intelligence agents are required to learn from their surroundings and to reason about the knowledge that has been learned in order to make decisions. While state-of-the-art deep learning typically uses distributed representations, reasoning is normally useful at a higher level of abstraction. As a result, attempts at combining symbolic AI and neural computation into neural-symbolic systems have been on the increase.
Neural-symbolic learning and reasoning is an area of research that aims at integrating relational/symbolic reasoning and knowledge representation (e.g., first order logic language for knowledge representation) by embedding their constructs in deep neural networks.
Image interpretation tasks, such as object detection, image classification or visual relationship detection, may benefit from neural-symbolic integration in order to represent and exploit high-level information about objects and their relationship. The goal of this project is to design deep architectures combining convolutional neural networks with neural-symbolic components such as Logic Tensor Networks to solve various image interpretation tasks. One of the potential advantages of this emerging technique is the possibility to exploit prior information available in the form of semantic networks, e.g. WordNet or ConceptNet, to improve the learning process.
Multiple thesis proposals are available tackling the following problems: i) techniques for end-to-end training of deep neuro-symbolic networks, ii) integration of prior knowledge in the training process, and iii) characterization of neural-symbolic architectures, e.g. in terms of types of features learnt, efficiency and interpretability. Skills possessed or to be acquired: programming skills (Python, Keras/Tensorflow, or other deep learning framework); good analytical and mathematical skills

Suggested readings:
https://arxiv.org/pdf/1705.08968
https://arxiv.org/pdf/2012.13635

See also  http://grains.polito.it/work.php


Deadline 09/02/2023      PROPONI LA TUA CANDIDATURA




© Politecnico di Torino
Corso Duca degli Abruzzi, 24 - 10129 Torino, ITALY
Contatti