How to build better models on diverse, high-quality data
Hey there! Strap in for another thrill ride through the federated frontier—where we're all about unlocking the coolest, crunchiest bits of secure, private ML magic.
What’s ahead in this newsletter:
Learning Hub: On drug discovery and secure data collaboration
News from Apheris: Apheris partners with AWS, highlights from Bits and Pretzels, and celebrating team achievements
Federated Minds Guest Section: How open-source innovations and the strategic use of proprietary data change your AI game – by CEO and co-founder of Apheris, Robin R?hm
Industry Round-up: Federated learning insights and applications in healthcare
Enjoy reading!
Learning Hub
The AI structural biology consortium: breaking data silos in pharma
After months of relentless effort, on a memorable Wednesday afternoon, we received the exhilarating news—we're officially powering the AI Structural Biology Consortium (AISB), alongside pharma giants Abbvie and Johnson & Johnson, and the world’s leading minds in model development.
But this isn't just a milestone for us — it's a transformative journey in drug discovery! In this consortium, pharma leaders are coming together to collaboratively train the latest foundational AI models for drug discovery on their proprietary protein structure data, all while upholding the strictest standards of data confidentiality
Introducing Apheris Compute Gateway 3.0
It's not every day that something as exciting as the Apheris Compute Gateway 3.0 launches. While some may get excited about the latest TV series, for us, it’s all about advancing secure and efficient data collaboration.
The newest version of our Compute Gateway confronts the challenges of compliance, data security, and privacy head-on. By integrating federated learning technology and focussing intensely on Governance, Security, and Privacy (GSP), we ensure that data collaborations are secure and compliant.
Check out the blog post by my colleague, Jan Stücke , to discover the full capabilities of the Apheris Compute Gateway 3.0.
News from Apheris
Apheris partners with AWS to improve drug discovery and development outcomes via federated learning
We're teaming up with Amazon Web Services (AWS) to enhance drug discovery through federated learning. This partnership ensures secure and private access to distributed drug discovery data.
Apheris is now available through AWSmarketplace
Thought working for a tech startup meant sitting behind a desk all day? Think again! Here are our two masterminds who launched the Apheris Compute Gateway on the AWS Marketplace last week. They're standing proudly in front of their achievement - because when you've just made it easy for the world to securely collaborate on data, it's hard not to smile.
Breaking news about AISB again?
Okay, I know you've heard me rave about it before, but I can't help but bring up the AI in Structural Biology Consortium (AISB) again. This time it's to give you a direct gateway to explore this innovative project in more detail on its own new website.
Find out how the Apheris Compute Gateway is securely orchestrating some of the world's most valuable proprietary datasets to develop more accurate and efficient molecular design models.
领英推荐
Apheris at Bits and Pretzels Conference
We enjoyed a fantastic day at Bits and Pretzels in Munich.
Here are our highlights:
There is a clear, visible shift toward more collaborative and secure approaches in healthcare data handling!
Celebrating our culture of excellence at Apheris
Big cheers for JT! He just received his Certified Artificial Intelligence Governance Professional (AIGP) credential from the International Association of Privacy Professionals (IAPP).
At Apheris, we're passionate about fostering a culture that empowers our team to explore, learn, and excel. Cheers to a culture that supports personal and professional growth!
Guest section: with Robin R?hm, CEO and co-founder of Apheris
This month, join Robin Roehm , Co-founder and CEO of Apheris, as he explores the thrilling advancements shaping AI across industries. With AI set to become a multi-trillion-dollar powerhouse by 2030, Robin breaks down how open-source innovations and the strategic use of proprietary data are game changers.
“Proprietary and distributed data has the potential to massively improve current AI models. We are seeing a spectacular acceleration of AI, transforming all sectors and creating a new multi-trillion-dollar industry by 2030. The widespread adoption of open-source model libraries shows that AI models have become rapidly commoditized, and a foundational model paradigm has emerged.
GPU cost and limited availability means that there is a rise in demand for smaller models customized on more tailored datasets. Hence, access to proprietary data is emerging as the key to differentiation for most enterprises looking to use AI to solve business problems.
The potential here extends far beyond language models. For example, Healthcare AI is becoming a $180B dollar industry. The rich set of multi-modal healthcare data – genomics, clinical, and images – has the potential to radically transform drug discovery, clinical trials and patient outcomes. Thousands of new healthcare AI companies will build specialized models – and they need access to proprietary data to do so.
We at Apheris help to safely connect distributed data for analytics and AI across boundaries. That's the mission we're trying to solve, and I'm really excited to see customers and partners creating real business value with our product.” Robin R?hm, Apheris
Industry round-up
Latest in federated learning
We found a comprehensive paper that rigorously tests federated learning against local models, demonstrating superior performance. Beyond the benefits, this paper offers an in-depth look at the real-world hurdles of FL projects, from legal to infrastructural challenges, with detailed insights and step-by-step guides on overcoming them.
This paper is a must-read for anyone interested in the practical implementation of federated learning, providing a roadmap for navigating the complexities involved.
A systematic review of federated learning in healthcare
With healthcare data making up about 30% of the global data volume, FL's ability to maintain patient privacy while enhancing model accuracy is more crucial than ever.
Explore the "Cambridge Mathematics of Information in Healthcare" paper, which delves into how federated learning (FL) can overcome major healthcare challenges such as isolated data sets and privacy concerns. This review offers a wealth of novel FL methodologies developed between 2015 and 2023, providing critical insights for advancing FL applications in healthcare.
Exploring LxMs in Biological Intelligence
"Generating or getting access to real-world multimodal datasets will be the most critical competitive moat for startups developing in the BioAI space" Vic Singh
Foundation models, or LxMs, are revolutionizing fields from genomics to drug discovery. Discover the advancements and companies leading the charge.
Cheers everyone! Marie Roehm