Learning for Structured Data

Tutorial @ ACDL 2020

16th-17th July 2020 - Certosa di Pontignano - Siena (IT)

Structured data are an effective representation for complex information, providing a straightforward means to bridge numerical data and symbolic relationships. The tutorial will provide an easy paced introduction to the lively field of deep learning for structured data and its applications, covering from foundational models to open research challenges. Dealing with structured information, such as trees and graphs, requires learning models capable of adapting to structured samples of varying size and topology, capturing the relevant structural patterns to perform predictive and explorative tasks while maintaining the efficiency and scalability necessary to process large scale networks. The tutorial will first introduce foundational aspects and seminal models for learning with structured data (trees, DAGs, graphs). Then it will discuss the most recent advancements in terms of deep learning for structured data processing, including learning structure embeddings, graph convolutions, attentional models. Finally, we will delve into open research challenges, applications, and interesting directions of future research.

Content & Materials

Syllabus

LECTURE 1

  • Motivations

  • Structured data types: sequences, trees and graphs

  • Formalization of the learning task: prediction, induction, transduction and generation with structured samples

  • Historical perspective: recurrent, recursive, contractive and contextual models

  • Learning with tree-structured data

    • Recursive gated networks

    • Generative and Bayesian models for trees

    • Tensor models

LECTURE 2

  • Learning with graph-structured data

    • Graph convolutions

    • Node embeddings

    • Unsupervised learning on graphs

LECTURE 3

  • Relevant current research challenges

    • Graph pooling: global, topological and adaptive methods

    • Graph generation: adjacency-based, language-based

    • Expressivity and theoretical aspects

    • Benchmarks and robust assessment of models

    • Software

    • Explainability

  • Applications

Slides

Tutorial slides are available here.

Bibliography

D. Bacciu, F.Errica, A. Micheli, M. Podda, A Gentle Introduction to Deep Learning for Graphs, Neural Networks, Vol 129, pp 203-221, 2020

Lecturer Biography

Davide Bacciu - UniversitĂ  di Pisa (bacciu@di.unipi.it)

Associate Professor at the Computer Science Department, University of Pisa. The core of his research is on Machine Learning (ML) and deep learning models for structured data processing, including sequences, trees and graphs. He is the PI of an Italian National project on ML for structured data and the Coordinator of the H2020-RIA project TEACHING (2020-2022). He has been teaching courses of Artificial Intelligence (AI) and ML at undergraduate and graduate levels since 2010. He is currently supervising 13 Ph.D. students in Computer Science and Data Science, including 5 Ph.D. students whose main focus is on learning for structured data. He is an IEEE Senior Member, the founder and chair of the IEEE Task Force on learning for structured data, a member of the IEEE NN Technical Committee and of the IEEE CIS Task Force on Deep Learning. He is an Associate Editor of the IEEE Transactions on Neural Networks and Learning Systems. Since 2017 he is the Secretary of the Italian Association for Artificial Intelligence (AI*IA).

Site under development. More information out soon.