APEX-AI 2027 — Call for Contributions

Approximate, Low-Precision, Sparsity, and Efficient AI Execution from Training to Deployment

January 19, 2027 | Glasgow, United Kingdom

Co-located with HiPEAC 2027

Workshop website: https://apex-ai-workshop.github.io/hipeac27/

About the Workshop

APEX-AI 2027 invites contributions on approximate and efficiency-oriented techniques for next-generation AI systems.

As AI models continue to grow in size and complexity, improving their efficiency increasingly requires coordinated advances across algorithms, software, and hardware. Techniques such as low- and adaptive-precision computing, sparsity, conditional execution, memory optimization, and hardware–software co-design are becoming essential to enable scalable and energy-efficient AI training and deployment.

APEX-AI aims to bring together researchers and practitioners from computer architecture, systems, machine learning, and design automation to discuss emerging techniques, practical challenges, and research opportunities in efficient AI execution.

The workshop will feature invited talks from academia and industry, complemented by a selected number of contributed technical presentations.

Topics of Interest

We welcome contributions addressing, but not limited to, the following topics:

Low-Precision and Approximate AI

• Low-precision arithmetic and representations (FP8, FP4, INT4, and beyond)

• Adaptive and mixed-precision computing

• Approximate arithmetic and accuracy–efficiency trade-offs

Sparsity and Conditional Execution

• Structured and unstructured sparsity

• Mixture-of-Experts (MoE) and dynamic routing

• Token, layer, and computation skipping

• Dynamic and input-dependent execution

Memory and Data Movement

• KV-cache optimization and compression

• Memory-efficient model execution

• Sparse memory access and data movement optimization

• High-bandwidth memory and interconnect efficiency

Efficient Large-Model Execution

• Efficient LLM training and inference

• Speculative decoding and inference acceleration

• Long-context processing

• Distributed and disaggregated inference

Accelerator–Compiler–Runtime Co-Design

• Domain-specific and heterogeneous AI acceleration

• Compiler and runtime optimizations

• Scheduling and resource management

• Cross-layer hardware–software optimization

Submission Guidelines

We invite one-page abstracts proposing technical presentations relevant to the workshop topics.

Submissions may describe:

• New or ongoing research

• Work in progress and preliminary results

• Previously published or presented research

• Practical experiences and lessons learned

• Emerging challenges and open research questions

• Technical perspectives and promising research directions

Abstracts must be written in English and submitted as a one-page PDF, including the proposed presentation title, presenter name(s), affiliation(s), and contact email.

Submissions should clearly communicate the main technical message, relevance to APEX-AI, and expected value for workshop discussions.

APEX-AI is a non-archival workshop. No formal proceedings are planned, and accepted abstracts will be selected for presentation only. Previously published work is therefore welcome.

Selected contributors will be invited to deliver a 15–20 minute technical presentation, followed by discussion.

Submissions will be evaluated based on relevance, technical interest, potential for stimulating discussion, and overall program balance.

At least one presenter of each accepted contribution must register for HiPEAC 2027 and attend the workshop in person.

Important Dates

• Abstract submission deadline: November 15, 2026 (AoE)

• Selection notification: December 8, 2026

• Final workshop schedule: December 15, 2026

• HiPEAC early registration: December 25, 2026

• Workshop date: January 19, 2027

Workshop Organizers

• Henk Corporaal — Eindhoven University of Technology (Chair)

• Stefano Corda — Huawei Europe (Co-Chair)

• Dimitrios Soudris — National Technical University of Athens

• Akash Kumar — Ruhr University Bochum

• Philipp Petersen — University of Vienna

• Sajjad Tamimi — Huawei Europe

• Zheng Li — Huawei Europe

Further Information

Workshop website: https://apex-ai-workshop.github.io/hipeac27/

APEX-AI 2027 is held in conjunction with HiPEAC 2027 in Glasgow, United Kingdom.