On September 20, Facebook hosted the first AI Systems Faculty Summit for 40 academic researchers and PhD students from 14 universities around the world. At the summit, AI Infrastructure Engineering Manager Kim Hazelwood announced the Systems for Machine Learning request for proposals (ML RFP), a new research award opportunity with the goal of fostering further innovation and deepening our collaboration with academia.
“We’re excited to announce our third RFP in the systems for ML space,” says Hazelwood. “As always, we’re interested in foundational advances in systems and developer efficiency for machine learning workloads, including content understanding and recommendation systems. New this year is our additional interest in systems support for privacy-preserving machine learning and emerging technologies.”
This year, we are particularly interested in proposals that fall into these categories:
- Scalable, elastic, and reliable distributed ML and inference
- System and architecture support for personalized recommendation systems
- Programming language and compilers for platform-agnostic ML
- Resource provisioning for efficient inference/training in heterogeneous data centers
- On-device training and inference
- System and architecture support for privacy-preserving ML
- System support for multiparty compute and private/secure inference
- Emerging technologies, such as near-memory processing and in-memory computing systems applied to ML
- Novel ML systems beyond neural networks
- Emerging technologies for efficient ML
“We look forward to receiving proposals that outline impactful new ways of applying computer systems research to modern applications of machine learning,” says Hazelwood.
In addition to presentations by several AI Infrastructure team members, this year’s AI Systems Faculty Summit featured lightning talks from the winners of the 2019 AI System Hardware/Software Co-Design RFP and the 2018 Hardware and Software Systems RFP.
Research Program Manager Sharon Ayalde explains the importance of continued collaboration with the academic community: “Bringing together the winners of our 2018 and 2019 requests for proposals allows us to continue the conversation: What challenges did we have last year, or even earlier this year, that we can implement solutions for or discuss further?” she says. “It’s important to understand how these winning projects have evolved and what new challenges we can seek expertise in — within this community and within the wider academic community.”
Following this model, we also plan to invite the future winners of the new Systems for ML research awards to present their findings in a 2020 summit. “Academia is great at tackling and solving hard fundamental problems and gaining deep insights,” says Misha Smelyanskiy, Director, AI System SW/HW Co-design group. “We have many such real production problems at Facebook that we face every day, and we are happy to engage with academia on solving them. By partnering together we can scale our impact, both figuratively and literally.”
For application requirements, eligibility, timing, and more, visit the Systems for ML RFP page.