Embeddings are one of the most versatile techniques in machine learning, and a critical tool every ML engineer should have in their tool belt. It’s a shame, then, that so few of us understand what they are and what they’re good for! The problem, maybe, is that embeddings sound slightly abstract and esoteric: In machine learning, an embedding is a way of representing data as points in n-dimensional space so that similar data points cluster together. Sound boring and unimpressive? Don’t be fooled. Because once you understand this ML multitool, you’ll be able to build everything from search engin ..read more

@azusagakuyuki is a young Japanese motorcyclist with long hair, a delicate chin, and 33,000 Twitter followers. There, she posts pictures of herself in a biker shirt, posing in front of her gleaming red-and-blue Yamaha Telkor on dirt roads and hilltops and misty beaches. She’s beautiful, adventurous, and envy-inducing.  But one day, she accidentally posted a picture of her bike on Twitter that captured her reflection in the rear-view mirror. The reflection was of a middle-aged man–because the woman in the photo was actually a 50-year-old man named Soya who transformed his face using a mach ..read more

Your manager holds a gun to the side of your head and says, “Sell me a Google Cloud AI product in the next ten minutes or you’re toast.” It’s that time of year again. Performance reviews. Perf. “Okay, okay,” you say. “Cool it. There are so many products in the GCP AI portfolio, I’m sure we can find one to fit your use case.” “I don’t have a use case,” your manager says. “AWS said I didn’t need one.” Now you’re sweating bullets. “And if you don’t start selling me something in the next three seconds, I’m turning this thing on,” he says, patting the metal rail next to you. You’re chained to a tre ..read more

You are dying. The reason why doesn’t matter. Maybe it’s that you spent all your best years with your nose to the grindstone, burning that midnight oil, ha ha. Or maybe it’s that, for the past fifteen years, you’ve subsisted exclusively on coffee and Soylent. Complete Nutrition Backed By Science, they said. No way to prove them wrong! Ha ha ha! Or maybe it was those fumes. You are going to give yourself immortal life. No–you are going to create a new, better version of yourself that’s immortal–a living replica of you made of metal that will act and say the things you would, if you were still a ..read more

Introducing PetCam: a non-invasive machine-learning-powered pet tracker that runs on an old smartphone. This project is a collaboration between me and Jason Mayes, who came up with the idea. Also, funny story, uh… my colleague Markku Lepistö built (almost) THE EXACT SAME PROJECT at the same time on his own YouTube show, Level Up, which you can see here. We use old smartphones. He uses a Coral development board. Choose your own adventure. When I was young and lived at home in New Jersey, my parents were really strict with me about remembering to close the garage at night. Because if I didn’t cl ..read more

You know that expression When you have a hammer, everything looks like a nail? Well, in machine learning, it seems like we really have discovered a magical hammer for which everything is, in fact, a nail, and they’re called Transformers. Transformers are models that can be designed to translate text, write poems and op eds, and even generate computer code. In fact, lots of the amazing research I write about on daleonai.com is built on Transformers, like AlphaFold 2, the model that predicts the structures of proteins from their genetic sequences, as well as powerful natural language processing ..read more

Let me start this post off by saying that imposter syndrome has already been covered profusely and at length, and there’s probably nothing new I can add to the discussion, so let me stop here, thanks for reading, and sorry for wasting your time. Akhem. While there’s already tons of advice for overcoming imposter syndrome, I find it usually falls into one of two buckets: YOU! An imposter?! No way! Just stop thinking that! Fake it ‘til you make it. If you just keep acting confident, one day you will be. The first angle is clearly useless, and the second, I’d argue, is neither possible nor advi ..read more

Alongside cooking for myself and walking laps around the house, anime (i.e. Japanese cartoons) is something I’ve learned to love during quarantine. The problem with watching anime, though, is that short of learning Japanese, you become dependent on human translators and voice actors to port the content to your language. Sometimes you get the subtitles (“subs”) but not the voicing (“dubs”). Other times, entire seasons of shows aren’t translated at all, and you’re left on the edge of your seat with only Wikipedia summaries and 90s web forums to ferry you through the darkness.  So what are y ..read more

Is it just this massive cold brew I’m drinking, or is OpenAI’s new image-generating model, DALL·E, really out of this world? This behemoth 12-billion-parameter neural network takes a text caption (i.e. “an armchair in the shape of an avocado”) and generates images to match it: From https://openai.com/blog/dall-e/. In July, the same company, OpenAI, released a similarly huge model called GPT-3 that wowed the world with its ability to generate human-like text, including Op Eds, poems, sonnets, and even computer code. Now they’ve applied the same technique–training a kind of neural network calle ..read more

At the end of last month, DeepMind, Google’s machine learning research branch known for building bots that beat world champions at Go and StarCraft II, hit a new benchmark: accurately predicting the structure of proteins. If their results are as good as the team claims, their model, AlphaFold, could be a major boon for both drug discovery and fundamental biological research. But how does this new neural-network-based model work? In this post, I’ll try to give you a brief but semi-deep dive behind both the machine learning and biology that power this model. First, a quick biology primer: The fu ..read more