TY - JOUR
T1 - DimReader
T2 - Axis lines that explain non-linear projections
AU - Faust, Rebecca
AU - Glickenstein, David
AU - Scheidegger, Carlos
N1 - Funding Information:
We acknowledge fruitful discussions of dimensionality reduction with Sean Stephens, Luiz Gustavo Nonato, and Joshua Levine on the energy formulation of t-SNE. This work has been partially supported by the NSF under the TRIPODS program, award number CCF-1740858, and IIS-1513651.
Publisher Copyright:
© 2018 IEEE.
PY - 2019/1
Y1 - 2019/1
N2 - Non-linear dimensionality reduction (NDR) methods such as LLE and t-SNE are popular with visualization researchers and experienced data analysts, but present serious problems of interpretation. In this paper, we present DimReader, a technique that recovers readable axes from such techniques. DimReader is based on analyzing infinitesimal perturbations of the dataset with respect to variables of interest. The perturbations define exactly how we want to change each point in the original dataset and we measure the effect that these changes have on the projection. The recovered axes are in direct analogy with the axis lines (grid lines) of traditional scatterplots. We also present methods for discovering perturbations on the input data that change the projection the most. The calculation of the perturbations is efficient and easily integrated into programs written in modern programming languages. We present results of DimReader on a variety of NDR methods and datasets both synthetic and real-life, and show how it can be used to compare different NDR methods. Finally, we discuss limitations of our proposal and situations where further research is needed.
AB - Non-linear dimensionality reduction (NDR) methods such as LLE and t-SNE are popular with visualization researchers and experienced data analysts, but present serious problems of interpretation. In this paper, we present DimReader, a technique that recovers readable axes from such techniques. DimReader is based on analyzing infinitesimal perturbations of the dataset with respect to variables of interest. The perturbations define exactly how we want to change each point in the original dataset and we measure the effect that these changes have on the projection. The recovered axes are in direct analogy with the axis lines (grid lines) of traditional scatterplots. We also present methods for discovering perturbations on the input data that change the projection the most. The calculation of the perturbations is efficient and easily integrated into programs written in modern programming languages. We present results of DimReader on a variety of NDR methods and datasets both synthetic and real-life, and show how it can be used to compare different NDR methods. Finally, we discuss limitations of our proposal and situations where further research is needed.
KW - Non-linear dimensionality reduction
KW - auto-differentiation
UR - http://www.scopus.com/inward/record.url?scp=85052654810&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85052654810&partnerID=8YFLogxK
U2 - 10.1109/TVCG.2018.2865194
DO - 10.1109/TVCG.2018.2865194
M3 - Article
AN - SCOPUS:85052654810
SN - 1077-2626
VL - 25
SP - 481
EP - 490
JO - IEEE Transactions on Visualization and Computer Graphics
JF - IEEE Transactions on Visualization and Computer Graphics
IS - 1
M1 - 8440820
ER -