Expanding Applications for ML Through Research
As machine learning (ML) expands to more applications across all areas of compute and the wider technology agenda, our research continues to guide and inform this growth. Arm advanced hardware, software, and tools provide the energy efficiency and performance required to support increasingly complex algorithms in this rapidly evolving area.
Key Research Threads
Our research covers a wide range of topics that focus on developing the technology to power future machine learning solutions.
Podcast: AI in Business
Our researchers share Arm’s latest cutting-edge ML research at top-tier conferences and events.
Meet the Team
Senior Director of Machine Learning
Matthew Mattina is head of Arm’s Machine Learning Research Lab, where he leads a team of world-class researchers developing advanced hardware, software, and algorithms for machine learning.
Join the team! We are always looking for talented researchers across all areas of ML. In particular, we are keen to hear from experts in probabilistic ML, including Bayesian inference, Gaussian processes, variational inference, probabilistic models, and ensemble learning.
Latest ML Research Blogs
Read more blogs on our community website.
Ensuring Your AI is Sure, Anywhere, Anytime
When developing ML applications, it’s important to define what we see and how well we it. Stochastic-YOLO adapts YOLOv3 architecture to generate uncertainty estimations by introducing stochasticity in the form of Monte Carlo Dropout, all while keeping efficiency in mind.
Using Multiple Labels Improves Neural Network Learning
Consider the problem of historical image ranking with the goal of accurately predicting the decade an image was taken. We use a standard classification loss function, while exploiting the ordinal information of the labels to classify them.
Efficient Bug Discovery for Hardware Verification
To design a machine containing no bugs, we must test every aspect, but even with a single one-second test, it would still take 1022 years. We use ML to efficiently identify bugs and see a 25% increase in efficiency over default verification workflow.