Visualizing Data Using t-SNE
Channel:Google TechTalks · Length:55:10 · Published:2013-11-07 · Views:203669
Summary
A Google Tech Talk by Laurens van der Maaten, co-creator of t-SNE, on visualizing high-dimensional data by mapping it to two dimensions. It explains how the technique works and how it compares with other visualization approaches.
Why HN recommends it
HN people point to it as the canonical explanation from the method's author whenever t-SNE comes up.
there's a really interesting Google techtalk on how it works — HN, 2013
Machine learning & deep learning · Data visualization & information design
Subtitles (excerpt)
GEOFFREY HINTON: I'd like to introduce Laurens van der Maaten, who is our speaker today. He's going to talk about a new visualization technique that works better than any other technique for laying out large data sets in 2D maps, so you can see the structure of them. Laurens got his Ph.D. in computer vision from Tilburg, as we say in English, in 2009. He did a post-doc at UCSD with Lawrence Saul in machine learning. He's now an assistant professor at Delft University of Technology. In 2008, he visited Toronto for six months, where we worked together on developing t-SNE. I'm very part one of my contributions, which was to think outside the box and say it would work better if we made the probabilities of our set of mutually exclusive events add up to four. And Laurens now tells me it works better if we make them add up to 12. OK, Laurens. [APPLAUSE] LAURENS VAN DER MAATEN: Thank you very much, Geoff. Thanks for having me. It's great to be here. So I'll talk today about the work I've been doing with Geoff that have continuing since I visited in Toronto on visualizing data using t-SNE. So the setting we consider in this talk is we're given a bunch of high dimensional objects, in this case a book called x1 to xn.
On Hacker News
Barnes-Hut uses a quadtree, doesn't it? I don't know whether the code was adapted to use an octree in 3D instead; maybe it was? FWIW there's a really interesting Google techtalk on how it works: [VIDEO] — a1k0n, 2013-11
Use non spatial dimensions as well, like color, size, and shape. You can also use clever projections like t-sne to reduce the number of dimensions while retaining some of the structure of the data ( [VIDEO] ) — vankessel, 2019-08
Thanks for the offer. I've seen this [VIDEO] and I think I understand how it works but still think the results are amazing. — zoren, 2016-11
Related videos
- The spelled-out intro to neural networks and backpropagation: building micrograd · Andrej Karpathy · 2:25:52
- But what is a neural network? | Deep learning chapter 1 · 3Blue1Brown · 18:40
- A Better Default Colormap for Matplotlib | SciPy 2015 | Nathaniel Smith and Stéfan van der Walt · Enthought · 19:08
- Hans Rosling's 200 Countries, 200 Years, 4 Minutes - The Joy of Stats - BBC · BBC · 4:47
- AlphaGo - The Movie | Full award-winning documentary · Google DeepMind · 1:30:28
- Transformers, the tech behind LLMs | Deep Learning Chapter 5 · 3Blue1Brown · 27:14
- USENIX Security '18-Q: Why Do Keynote Speakers Keep Suggesting That Improving Security Is Possible? · USENIX · 51:22
- Lecture 1 | Machine Learning (Stanford) · Stanford · 1:08:40
- MarI/O - Machine Learning for Video Games · SethBling · 5:57
- Deep Dive into LLMs like ChatGPT · Andrej Karpathy · 3:31:23
- Computer program that learns to play classic NES games · suckerpinch · 16:17
- RL Course by David Silver - Lecture 1: Introduction to Reinforcement Learning · Google DeepMind · 1:28:12
- Let's build GPT: from scratch, in code, spelled out. · Andrej Karpathy · 1:56:20
- Bay Area Vision Meeting: Unsupervised Feature Learning and Deep Learning · Google TechTalks · 48:10
- A Hackers' Guide to Language Models · Jeremy Howard · 1:31:13
- Wealth Inequality in America · politizane · 6:23