Nvidia Patents a Smarter Batching Method for Graph Neural Network Training
Training graph neural networks is notoriously wasteful — most batching methods ignore the graph's own structure, so each training step spends half…
The Big Tech patents worth knowing in ai/ml, in plain English. Updated weekly from new USPTO publications.
Training graph neural networks is notoriously wasteful — most batching methods ignore the graph's own structure, so each training step spends half…
Teaching a robot to do something useful requires enormous amounts of demonstration data — and collecting that data by hand is slow,…
When Google's AI gives you an answer, how does it decide which sources to pull from? A new patent reveals that relevance…
Dropping a new object into a photo sounds simple — until the AI places a car on the ceiling. Nvidia's new patent…
Google published its Wide & Deep learning architecture in a 2016 research paper — and is now, apparently, patenting it. That gap…
Not all LinkedIn connections are equal — some are close colleagues, others are people you met once at a conference. Microsoft is…
Training large language models on very long documents is expensive — partly because the math that lets tokens 'attend' to each other…
Training large AI models on very long sequences of text is brutally expensive — partly because GPUs end up waiting around with…
Training a single AI model to understand dozens of languages is hard enough — doing it when those languages use completely different…
What if you could type one message — 'Make a Father's Day card using a photo of Viktor and Hannah' — and…
Google is patenting a way to dramatically shrink your stored photos — and then reconstruct the originals later using a personalized AI…
Anyone who's watched a dubbed foreign film knows the lip-sync horror — mouths moving while the words have already stopped. Microsoft is…
We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.