CodeGNN

Abstract

CodeGNN is a collaborative project supported by three research labs namely:

Laboratory	Supporting institution	Towns
		Caen
		Rouen and Le Havre
		Tours

These three labs aim to investigate Graph convolution networks in order to solve 3 important bottlenecks:

1. Convolution on graphs correspond mainly to low pass filters. We plan to propose new convolution schemes extending the range of available/learnable frequencies together with the items (sub graphs or vertices) on which these convolutions are applied.
2. Few methods of graph downsampling have been proposed so far. We plan to desing new downsampling/pooling methods allowing a fixed decimation ratio, a better control of the extend of each clusters and a preservation of the spectral properties of graphs.
3. Designing convolutional Graph neural network is not a so easy task. The problem becomes even more difficult when one consider sequences of graphs instead of graph instances. We plan to investigate these types of objects in order to propose new integration schemes of the temporal and spatial (graph) information.

These three research lines come with three applications conducted in close collaborations with three external research laboratories:

1. Investigating chemical properties using new convolution schemes and the electronic properties of atoms in close collaboration with :
   
2. Learning automatically the best clustering of the atoms of a molecule in order to predict its biological effect. We would like thus to affine thanks to a learning scheme, the pharmacophoric graphs used by biologists when designing a new drug.
   
3. Analysing sequences of fMRI acquisitions and sequences of body positions using techniques derived from the the analysis of graph sequences.
   

Context

A large amount of the recent advances in Artificial Intelligence may be connected to the Statistical pattern recognition field. Using continuous optimisation, this research field defines a global numerical description of objects that is then combined with machine learning techniques. On the other hand, most of the objects of interest of our today life are based on discrete objects with sequential (strings) or more complex (graph) relationships. We can evoke among many others, the relationships between people in social graphs, the bounds between atoms in a molecule or the topographic distance between speed sensors in traffic analysis, to name a few.

The prediction of the properties of such objects falls in the scope of structural pattern recognition. For decades, this research field has been limited by costly (e.g. based on subgraph isomorphism) or poorly efficient metrics usually combined with limited machine learning algorithms (mainly k-nearest neighbours algorithm). A first important breakthrough has been achieved by the introduction of structural kernels dedicated to strings or graphs. In addition to provide efficient metrics on these discrete objects, the latter are gateway towards many machine learning methods; Hence, they reduce the gap between structural and statistical pattern recognition techniques.

A second breakthrough in this field has been provided by the introduction of Graph Neural Networks (GNNs). As graph kernels, these networks provide a strong connection between graphs and machine leaning techniques. Moreover, as other deep learning techniques, GNNs avoid handcrafting the design of a similarity measure between graphs. GNNs are based on two operations, namely: Graph convolution and Graph decimation/pooling. WP1 and WP2 are dedicated to overcome current limitations on these two operations. WP1 will investigate the design of more selective convolution filters in the spectral domain while WP2 will consider alternatives to clustering as a base for a Graph decimation. Dynamicity of time-evolving graphs constitutes a key concern in many practical applications, but is much less investigated. The point here is to perform a prediction about graph sequences instead of single graphs.

Applications concern, among others, the analysis of Brain’s graph sequences, traffic forecasting, action recognition . . . . This point will be addressed by WP3. This work package will combine the findings of WP1 and WP2 while integrating the time dimension. In close collaboration with brain specialists, this package will target the prediction of waking and falling asleep phases in functional MRI. Such a prediction is fundamental for a correct interpretation of functional MRI sequences where patients and especially the elderly can go through different levels of consciousness

Consortium

The project groups 3 French research laboratories, namely the GREYC (UMR 6072), the LITIS EA 4108 and the LIFAT EA 6300 – ERL CNRS 7002.

The members of the project are listed bellow:

Partner	Name	First Name	Position	Role	Expertise in the project
	Brun	Luc	PR	Project coordinator WP leader WP2	Graph Decimation
	Cuissart	Bertrand	MCF		major: graph algorithmics, minor : aggregation of information
	Bougleux	Sébastien	MCF		major: Analysis of time-varying graphs
	Lechervy	Alexis	MCF		major: Machine learning, minor: Analysis of time sequences
	Stanovic	Stevan	PhD student GREYC/LITIS		major: Graph Reduction
	Adam	Sébastien	PR	Local Coordinator LITIS	major:Graph convolution, minor: Graph neural networks for spatio-temporal data
	Héroux	Pierre	MCF	WP leader WP1	major:Graph convolution
	Chatelain	Clément	MCF		major:Recurrent Neural Networks
	Gaüzère	Benoît	MCF		major:graph similarity measures minor:Graph convolutions
	Honeine	Paul	PR		major:Graph pre-image \& autoencoders minor:learning theory
	Yger	Florian	MCF* associated		major:Graph representation minor: NN on Riemannian manifolds
	Conte	Donatello	MCF	WP Leader WP3	major:graphs sequences analysis minor: GCNN
	Ramel	Jean-Yves	PR	Local Coordinator LIFAT	major:Graph Embedding minor: Graph based representations
	Raveaux	Romain	MCF		major:GCNN in graph space
	Ragot	Nicolas	MCF		major: Recurrent Neural Networks

Meetings

title	date	location
Kick Off meeting of CodeGNN	April 25, 2022	Rouen
Second meeting of CodeGNN	April 4, 2023	Tours
Third meeting of CodeGNN	December 8, 2023	Caen

Publications

• Luc Brun, Technical Report : Pooling properties within the Graph Neural network framework(2023), HAL-ID: hal-04379337.
• Jason Piquenot, Aldo Moscatelli, Maxime Bérar, Pierre Héroux, Jean-Yves Ramel, Romain Raveaux, Sébastien Adam: Technical report: Graph Neural Networks go Grammatical. CoRR abs/2303.01590 (2023)
• Stanovic, Stevan, Gauzere, Benoit & Brun, Luc. Maximal Independent Sets for Pooling in Graph Neural Networks. In Mario Vento, Pasquale Foggia, Donatello Conte and Vincenzo Carletti editors. , Graph-Based Representations in Pattern Recognition – 13th IAPR-TC-15 International Workshop, GbRPR 2023, Vietri sul Mare, Italy, September 6-8, 2023, Proceedings , pages 113–124 2023.
• Stanovic, Stevan, Gauzere, Benoit & Brun, Luc. Ensemble de sommets independant maximal applique au pooling sur graphes. In Congres Reconnaissance des Formes, Image, Apprentissage et Perception ( RFIAP) , VANNES, France , Jul 2022 .
• Stanovic, Stevan, Gauzere, Benoit & Brun, Luc. Maximal Independent Vertex Set applied to Graph Pooling. In Structural and Syntactic Pattern Recognition . In Proceedings of SSPR’2022 conference, Montreal, Canada , Aug 2022
• Nicolas boria, Jana Kiederle, Florian Yger, David B. Blumenthal. The edge-preservation similarity for comparing rooted, unordered, node-labeled trees. In Pattern Recognition Letters 167(), 189-195.