Apple researchers will present a series of important research papers at NeurIPS 2025, one of the world’s leading artificial intelligence (AI) and machine learning (ML) events, in December.
Make models great again
Apple’s paper hints at elements of the direction the company is taking ML and AI, primarily around efficiency and privacy. Apple’s research teams share their findings at NeurIPS to accelerate industry progress.
In a blog post on its machine learning website, Apple says it is particularly focused on three of the many papers that will be presented at the event.
- STARFlow focuses on a scalable approach that produces high-resolution images of comparable quality without the computational cost and complexity of traditional methods. “STARFlow produces images at resolution and quality levels previously thought unattainable with NF models, rivaling best-diffusion and autoregressive methods while maintaining accurate likelihood modeling and faster inference,” Apple said.
- “Illusions of Thinking” investigates how current AI models handle complex reasoning tasks. In this study, we systematically test how the performance of these models changes as problem complexity increases and find that the accuracy of frontier large-scale reasoning models (LRMs) collapses above a certain complexity.
- Instance optimality for private KL distribution estimation. We explore how to accurately estimate probability distributions while preserving privacy.
NeurIPS attendees can experience demonstrations of Apple’s ML research in booth 1103 during the exhibit. Apple also sponsors and participates in several events hosted by affinity groups that support underrepresented groups in the ML community.
huge industry event
A comprehensive overview of Apple’s participation and contributions to NeurIPS 2025 can be found here. Thanks to this handy online infographic, you can explore the vast (huge, in fact) collection of over 6,000 research papers that will be presented at the event.
Like the history of AI, NeurIPS goes back a long way. The first conference was held in 1987. Initially focused on biological and artificial neural networks, recent events have focused on machine learning and AI. Expect lectures, demonstrations, symposiums, oral and poster presentations of peer-reviewed papers, as well as tutorials and workshops.
Follow me on social media! Join us on BlueSky, LinkedIn, and Mastodon.
