Speaker Details

Guy Royse

Redis Labs

Guy works for Redis Labs as a Developer Advocate. Combining his decades of experience in writing software with a passion for sharing what he has learned, Guy goes out into developer communities and helps others build great software. Teaching and community have long been a focus for Guy. He is President of the Columbus JavaScript Users Group, an organizer for the Columbus Functional Programmers, and has even has helped teach programming at a prison in central Ohio. In his personal life, Guy is a hard-boiled geek interested in role-playing games, science fiction, and technology. He also has a slightly less geeky interest in history and linguistics. In his spare time he likes to camp and study history and linguistics. Guy lives in Ohio with his wife, his three teenage sons, and an entire wall of games.

Mystery Machine Learning: Classifying Text with Recurrent Neural Networks, Keras, and Scoob and the Gang

Jinkies! Spoiler Alert! If you’ve seen Scooby-Doo, you know who the monster always is—Old Man Withers, the guy who runs the amusement park. Amusement park operators like Old Man Withers have caused Mystery, Inc. all sorts of problems over the years. Problems that could be avoided with a properly trained recurrent neural network. With RNNs, Scoob and the Gang could have built a model to classify everyone’s speech. This would show them that Old Man Withers and Redbeard’s Ghost sounded a lot alike!

We can help Mystery, Inc. with this problem by building a recurrent neural network to do just that. You know we got a mystery to solve, and we’re going to solve it by building our model using lines from Scooby-Doo, Keras, and TensorFlow. Once we have our model, we’ll host it on RedisAI so we can quickly build an application to make use of it. Well, we’ve got some work to do now.

I will also explain what neural networks are in general, what recurrent neural networks are in particular, and discuss some practical use of this technology. When we’re done, you’ll know how to build RNNs with Keras, use them to classify text, and integrate them into your application. But more importantly, you're going to have yourself a Scooby snack! That’s a fact!

Artificial Intelligence
Neural Networks
Deep Learning

Understanding Probabilistic Data Structures with 112,092 UFO Sightings

If you’re like most developers, you probably have no idea what probabilistic data structures are. In fact, I did a super-scientific poll on Twitter and found that out of 119 participants, 58% had never heard of them and 22% had heard the term but nothing more. I wonder what percentage of that 22% heard the term for the first time in the poll. We’re a literal-minded lot at times.

Anyhow. That’s 4 out of 5 developers. And that’s a lot of folks that need to be educated. So, let’s do that.

A probabilistic data structure is, well, they’re sort of like the TARDIS—bigger on the inside—and JPEG compression—a bit lossy. And, like both, they are fast, accurate enough, and can take you to interesting places of adventure. That last one might not be something a JPEG does.

More technically speaking, most probabilistic data structures use hashes to give you faster and smaller data structures in exchange for precision. If you’ve got a mountain of data to process, this is super useful. In this talk, we’ll briefly go over some common probabilistic data structures; dive deep into a couple (Bloom Filter, MinHash, and Top-K); and show a running application that makes use of Top-K in Redis to analyze the most commonly used words in all 112,092 of my UFO sightings.

When we’re done, you’ll be ready to start using some of these structures in your own applications.

Data Structures