Machine Learning/CORE Unranked

AXIOM 2026

AXIOM: Foundations of Efficient Deep Learning

Sep 12 2026ParisOfficial conference site Site reachable

AXIOM 2026 is a NeurIPS workshop focused on establishing predictive principles for efficient deep learning by bridging theoretical advances with practical efficiency challenges. It brings together researchers in deep learning theory, optimization, systems, and hardware to address how mathematical foundations can guide the design of AI systems constrained by compute, memory, energy, and data.

Official CFP Back to deadlines Verified September 9, 2026

Key deadlines

Verified September 9, 2026

Full paper

August 29, 2026

AoE

Conference timeline

Submission and decisions

Full paperKey deadline

August 29, 2026 · AoE

Paper fit

Contribution paths

A strong submission should clearly identify its contribution and evaluate it appropriately.

Research Papers

Short papers (4 pages, excluding references) following NeurIPS workshop formatting guidelines; original research, preliminary results, novel ideas, or emerging directions advancing the theoretical and practical foundations of efficient deep learning.

Grand Challenges

One-page abstracts (excluding references, no appendix) articulating open theoretical questions, theoretical gaps, contradictions between theory and practice, or future research opportunities in efficient AI.

Research areas in scope

Topics of Interest

Predictive Theory for Efficient Learning: Scaling laws for efficient models, compute-optimal training and inference, predicting capability under resource constraints, optimization and learning dynamics, generalization under limited compute or data, phase transitions in efficient learningSparsity, Compression, and Model Structure: Foundations of pruning and quantization, sparse and modular neural networks, lottery tickets and subnetworks, adaptive computation, neural architecture design, representation learning for efficiencyEfficient Foundation Models: Efficient LLMs and multimodal models, efficient reasoning and adaptive inference, mixture-of-experts and modular architectures, test-time adaptation, memory-efficient training and inference, distillation and compressionFoundations and Future Directions: Theoretical limits of efficient AI, new efficiency metrics and benchmarks, predictive models of training dynamics, interpretability of efficient models, mathematical foundations of efficient deep learning, emerging theoretical paradigms for efficient AI

Policies worth checking twice

Submissions must be double-blind
Papers must be 4 pages excluding references and appendix
Grand Challenges submissions must be 1 page excluding references with no appendix
Use of NeurIPS 2026 LaTeX template is required
Workshop is non-archival
No duplication of work previously published at machine learning or related conferences
Work presented at the main NeurIPS conference cannot also appear in the workshop
All submissions undergo peer review by the Program Committee

Official sources

Compiled from the official call for papers. The organizers’ pages remain authoritative.

Official conference site

Last verified September 9, 2026