CellProfiler extracts ~ 100 - 1000 cell level features
We perform feature / dimensional reduction and calculate the mean for each perturbation
Word2Vec
DCGANs
+
=
Cells treated with A
Cells treated with B
Cells treated with A and B ?
or not?
Can we learn a robust representation from unlabeled data ?
1000-dim CP feature vector
https://swarbrickjones.files.wordpress.com
cropped cell images
latent space
unsupervised clustering in latent space
(semi) supervised training using hand-labeled data
Classifying and Segmenting Microscopy Images Using Convolutional Multiple Instance Learning, Arxiv 2015, O Kraus, LJ Ba, B Frey
Spatial Transformer Networks NIPS 2015 Max Jaderberg, Karen Simonyan, Andrew Zisserman, Koray Kavukcuoglu
By David Dao