Visualizing Data using t-SNE

← Back to document page

JournalofMachineLearningResearch9 ,5000LETilburg, King s College Road,M5S3G4Toronto,ON,CanadaEditor:Yoshu aBengioAbstractWe presenta newtechniquecalled t-SNE thatvisualizeshigh-dimensionaldatabygivi ngeachdatapointa locationina two a variationofStochasticNeighborEmbedding(H intonandRoweis,2002)thatismucheasiertoop timize,andproducessignificantlybettervis ualizationsbyreducingthetendency betterthanexistingtechniquesat creatinga singlemapthatrevealsstructureat many particularlyimportantforhigh-dimensional datathatlieonseveraldifferent,butrelated ,low-dimensionalmanifolds, ,weshow howt-SNEcanuserandomwalksonneighborhoodg raphstoallowtheimplicitstructureofalloft hedatato influencethewayin whicha subsetofthedatais illustratetheperformanceoft-SNEona wi

VAN DER MAATEN AND HINTON a small qjji to model a large pjji), but there is only a small cost for using nearby map points to represent widely separated datapoints. This small cost comes from wasting some of the probability mass in the relevant Q distributions. In other words, the SNE cost function focuses on retaining the

  Der van