Can we see the hidden patterns in a cloud of data points?
Image: Kathryn Tunyasuvunakool, Jonas Adler, Zachary Wu, Tim Green, Michal Zielinski, Augustin Žídek, Alex Bridgland, Andrew Co, CC BY 4.0, via Wikimedia Commons
t-SNE preserves local structure
Can we see the hidden patterns in a cloud of data points?
Imagine trying to understand the complex relationships in a social network graph, where each person is a data point and connections represent friendships.
t-SNE helps us see the social network by grouping friends closer together and strangers farther apart, making it easier to spot communities and connections.
Example
In a social network with 100 people, t-SNE might show us that friends often cluster together, while distant acquaintances are spread out.
Remember this
t-SNE transforms distances into probabilities, revealing the underlying structure in high-dimensional data.
Text adapted from Wikipedia, licensed under CC BY-SA 4.0.
Related concepts
the Johnson-Lindenstrauss lemma says
Can we shrink big data without losing important details?
PCA vs t-SNE: PCA preserves global variance linearly, t-SNE preserves local structure nonlinearly
How can we teach computers to understand what we like?
UMAP is faster than t-SNE
UMAP is faster due to approximate nearest neighbors and cross-entropy optimization
non-convex loss landscapes are hard: many local minima and saddle points
Why do some hills have more tricky paths than others?
random projection to O(log n/ε²) dimensions preserves pairwise distances within 1±ε
Can you shrink a 3D object into 2D without losing its shape?
cosine similarity is preferred over dot product for normalized embeddings
Why do we need a special way to measure similarity in high-dimensional spaces?
Swipe through 100 ML concepts daily
Open Pocket Polymath