home | office of planning and analysis...nov 06, 2017 · opportunities for rsrch ito produce...
TRANSCRIPT
•
•
•
•
•
•
•
•••
•
•
•
•
•
•
•
t-SNEt-SNE
Google Brain Team
https://skybluetrades.net/blog/posts/2011/10/30/machine-learning/test-swiss-roll.png
https://experiments.withgoogle.com/ai/visualizing-high-dimensional-space
Undergraduate students who graduated in Spring 2016,grouped by the lower division subjects that they tookat UC Berkeley.
(N = 3,782)freshmen entrants
Undergraduate students who graduated in Spring 2016,grouped by the upper division subjects that they tookAt UC Berkeley.
(N = 3,785)freshmen entrants
“Though [visualizing a two or three dimensional representation of high-dimensional data] may seem like a trivial point, many statistical and machine learning algorithms have very poor optimality guarantees, so the ability to actually see the data and the output of an algorithm is of great practical interest.”
- Prateek Joshi
Visualizations by the Office of Planning & Analysis
Here’s a list of some very helpful t-SNE resources:How to Use t-SNE EffectivelyAn illustrated introduction to the t-SNE algorithmVisualizing MNIST: An Exploration of Dimensionality ReductionWhat is Manifold Learning?