LoG 2026 — Learning on Graphs Conference

LoG 2026, the Learning on Graphs Conference, is a fully in-person research conference held November 20–22, 2026 at Northeastern University in Boston, Massachusetts. It is the venue dedicated to machine learning on graphs and geometry: expressive graph neural networks, equivariant architectures, geometric processing, graph generative models, knowledge graphs, molecular applications, recommender systems, self-supervised and spectral methods, and the growing overlap between graphs and large language models. Papers go through a proceedings track, published in the Proceedings of Machine Learning Research (PMLR) with up to nine pages, or a four-page extended-abstract track. The conference makes review quality an explicit focus, and it also runs open calls for tutorials and a network of regional LoG meetups around the world.

• Notification to authorsSep 13, 2026
• Camera-ready deadlineOct 15, 2026

Date

November 20–22, 2026

Location

Northeastern University, Boston, Massachusetts

Format

In-Person

Event Type

Conference

Status

Upcoming

Organizer

Learning on Graphs Conference

Added

Sep 11, 2026

Updated

Sep 11, 2026

• ✓Graph ML and GNN researchers
• ✓Knowledge graph and data platform architects
• ✓Recommender systems engineers
• ✓Computational chemistry and drug discovery ML teams
• ✓Applied scientists working on LLM + graph retrieval
• ✓Graduate students in machine learning

Knowledge graphs, graph-based retrieval for LLMs, fraud and recommendation networks, and molecular modelling all run on the methods presented at LoG. For a CTO or head of applied science, it is the most concentrated place to judge which graph neural network techniques are production-ready, which benchmarks hold up, and where combining graphs with LLMs actually improves grounding. Its small, in-person format makes it a practical place to recruit graph-ML specialists, who are scarce in the hiring market.

• ✓Proceedings track (up to 9 pages, published in PMLR)
• ✓Extended abstract track (up to 4 pages)
• ✓Call for tutorials
• ✓CFP area: expressive GNNs and equivariant architectures
• ✓CFP area: knowledge graphs, recommender systems and graphs + LLMs
• ✓CFP area: molecular and geometric applications

• ★The only conference dedicated entirely to graph and geometric machine learning, rather than a workshop at a general ML venue
• ★Makes review quality an explicit priority of the conference
• ★Proceedings published in PMLR, alongside a lighter four-page extended-abstract track
• ★Backed by a worldwide network of regional LoG meetups between editions

Registration fees have not yet been published on logconference.org.

When is LoG 2026?

LoG 2026 takes place November 20–22, 2026 in Boston, Massachusetts.

Where is LoG 2026 held?

LoG 2026 is held in Boston, Massachusetts. The event format is in-person.

What is the LoG 2026 submission deadline?

The next LoG 2026 deadline is notification to authors on Sep 13, 2026. Remaining dates: Camera-ready deadline — Oct 15, 2026.

Who should attend LoG 2026?

LoG 2026 is aimed at Graph ML and GNN researchers, Knowledge graph and data platform architects, Recommender systems engineers, Computational chemistry and drug discovery ML teams, Applied scientists working on LLM + graph retrieval, Graduate students in machine learning.

What topics does LoG 2026 cover?

LoG 2026 covers Graph neural networks, Equivariant and geometric deep learning, Knowledge graphs, Graph generative models, Molecular machine learning, Recommender systems, Graphs and large language models, Self-supervised and spectral graph methods.

This page was written from 3 sources, 1 on domains other than logconference.org.

1. 1.logconference.org — logconference.orgvendor
2. 2.logconference.org — cfpvendor
3. 3.iaifi.org — related events