Join the event trusted by enterprise leaders for nearly two decades. VB Transform brings together the people building real enterprise AI strategy.\u00a0Learn more<\/em><\/p>\n\n\n\n Apple\u2018s machine learning research team has developed a breakthrough AI system for generating high-resolution images that could challenge the dominance of diffusion models, the technology powering popular image generators like DALL-E and Midjourney.<\/p>\n\n\n\n The advancement, detailed in a research paper published last week, introduces \u201cSTARFlow,\u201d a system developed by Apple researchers in collaboration with academic partners that combines normalizing flows with autoregressive transformers to achieve what the team calls \u201ccompetitive performance\u201d with state-of-the-art diffusion models.<\/p>\n\n\n\n The breakthrough comes at a critical moment for Apple, which has faced mounting criticism over its struggles with artificial intelligence. At Monday\u2019s Worldwide Developers Conference, the company unveiled only modest AI updates to its Apple Intelligence platform, highlighting the competitive pressure facing a company that many view as falling behind in the AI arms race.<\/p>\n\n\n\n \u201cTo our knowledge, this work is the first successful demonstration of normalizing flows operating effectively at this scale and resolution,\u201d wrote the research team, which includes Apple machine learning researchers Jiatao Gu, Joshua M. Susskind, and Shuangfei Zhai, along with academic collaborators from institutions including UC Berkeley and Georgia Tech.<\/p>\n\n\n\n The STARFlow research represents Apple\u2019s broader effort to develop distinctive AI capabilities that could differentiate its products from competitors. While companies like Google and OpenAI have dominated headlines with their generative AI advances, Apple has been working on alternative approaches that could offer unique advantages.<\/p>\n\n\n\n The research team tackled a fundamental challenge in AI image generation: scaling normalizing flows to work effectively with high-resolution images. Normalizing flows, a type of generative model that learns to transform simple distributions into complex ones, have traditionally been overshadowed by diffusion models and generative adversarial networks in image synthesis applications.<\/p>\n\n\n\n \u201cSTARFlow achieves competitive performance in both class-conditional and text-conditional image generation tasks, approaching state-of-the-art diffusion models in sample quality,\u201d the researchers wrote, demonstrating the system\u2019s versatility across different types of image synthesis challenges.<\/p>\n\n\n\n Apple\u2019s research team introduced several key innovations to overcome the limitations of existing normalizing flow approaches. The system employs what researchers call a \u201cdeep-shallow design,\u201d using \u201ca deep Transformer block [that] captures most of the model representational capacity, complemented by a few shallow Transformer blocks that are computationally efficient yet substantially beneficial.\u201d<\/p>\n\n\n\n The breakthrough also involves operating in the \u201clatent space of pretrained autoencoders, which proves more effective than direct pixel-level modeling,\u201d according to the paper. This approach allows the model to work with compressed representations of images rather than raw pixel data, significantly improving efficiency.<\/p>\n\n\n\n Unlike diffusion models, which rely on iterative denoising processes, STARFlow maintains the mathematical properties of normalizing flows, enabling \u201cexact maximum likelihood training in continuous spaces without discretization.\u201d<\/p>\n\n\n\n The research arrives as Apple faces increasing pressure to demonstrate meaningful progress in artificial intelligence. A recent Bloomberg analysis highlighted how Apple Intelligence and Siri have struggled to compete with rivals, while Apple\u2019s modest announcements at WWDC this week underscored the company\u2019s challenges in the AI space.<\/p>\n\n\n\n For Apple, STARFlow\u2019s exact likelihood training could offer advantages in applications requiring precise control over generated content or in scenarios where understanding model uncertainty is critical for decision-making \u2014 potentially valuable for enterprise applications and on-device AI capabilities that Apple has emphasized.<\/p>\n\n\n\n The research demonstrates that alternative approaches to diffusion models can achieve comparable results, potentially opening new avenues for innovation that could play to Apple\u2019s strengths in hardware-software integration and on-device processing.<\/p>\n\n\n\n The research exemplifies Apple\u2019s strategy of collaborating with leading academic institutions to advance its AI capabilities. Co-author Tianrong Chen, a PhD student at Georgia Tech who interned with Apple\u2019s machine learning research team, brings expertise in stochastic optimal control and generative modeling.<\/p>\n\n\n\n The collaboration also includes Ruixiang Zhang from UC Berkeley\u2019s mathematics department and Laurent Dinh, a machine learning researcher known for pioneering work on flow-based models during his time at Google Brain and DeepMind.<\/p>\n\n\n\n \u201cCrucially, our model remains an end-to-end normalizing flow,\u201d the researchers emphasized, distinguishing their approach from hybrid methods that sacrifice mathematical tractability for improved performance.<\/p>\n\n\n\n The full research paper is available on arXiv, providing technical details for researchers and engineers looking to build upon this work in the competitive field of generative AI. While STARFlow represents a significant technical achievement, the real test will be whether Apple can translate such research breakthroughs into the kind of consumer-facing AI features that have made competitors like ChatGPT household names. For a company that once revolutionized entire industries with products like the iPhone, the question isn\u2019t whether Apple can innovate in AI \u2014 it\u2019s whether they can do it fast enough.<\/p>\n Daily insights on business use cases with VB Daily<\/strong><\/p>\n If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI.<\/p>\n \n\t\t\t\t\tThanks for subscribing. Check out more VB newsletters here.\n\t\t\t\t<\/p>\n An error occured.<\/p>\n<\/p><\/div>\n
\n<\/div>How Apple is fighting back against OpenAI and Google in the AI wars<\/h2>\n\n\n\n
Inside the mathematical breakthrough that powers Apple\u2019s new AI system<\/h2>\n\n\n\n
What STARFlow means for Apple\u2019s future iPhone and Mac products<\/h2>\n\n\n\n
Why Apple is betting on university partnerships to solve its AI problem<\/h2>\n\n\n\n
![\"\"\/](\"https:\/\/venturebeat.com\/wp-content\/themes\/vb-news\/brand\/img\/vb-daily-phone.png\")
\n\t\t\t\t<\/div>\n<\/p><\/div>\n<\/div>\t\t\t<\/div>\r\n