Measuring the Interpretability of Unsupervised Representations via Quantized Reverse Probing.

Iro Laina Yuki M. Asano Andrea Vedaldi

Published in: CoRR (2022)

Keyphrases

video sequences
image sequences
unsupervised learning
prediction accuracy
supervised learning
data driven
multiscale
symbolic representation
low level
topic modeling
semi supervised
unsupervised feature selection
dct coefficients
representation scheme
database
pairwise
object recognition
neural network
data sets