Testing Pytorch Auto-Encoder Code (Deep Learning)
Happy Thursday Everyone. I apologize in advance for the LONG video but I promise it's beneficial!
A while ago I played with the idea of using an autoencoder to embed spatial transcriptomic gene expression that might allow us to do some cool things and visualizations. However, in trying to use the CosMx SMI data the embedding never seemed to work. In this video I wanted to revisit this idea and make sure that the code I was using works on more structured data. Right now am hypothesizing that the SMI data is perhaps too noisy to effectively be embedded in such a latent space.
To make sure that this code can be used for spatial transcriptomic data in the future, I create a custom dataset with the MNIST data from CSV files rather than from the pytorch package. Then continue on with an existing example of autoencoder code with some minor modifications.
Please consider contributing to my Patreon where I may do merch, gather ideas for future content, and have further discussions: / alexsoupir
The previous cuda conda repo for getting everything set up:
https://github.com/ACSoupir/cuda_conda
Spyder!! Love this python IDE:
https://anaconda.org/anaconda/spyder
MNIST Dataset as CSV rather than premade python objects or within the pytorch package as a dataset - allows us to make sure when applying to other tabular data our dataset/dataloader works as expected:
https://www.kaggle.com/datasets/oddra...
The Autoencoder reference used:
https://www.geeksforgeeks.org/impleme...
Sur cette page du site, vous pouvez voir la vidéo en ligne Testing Pytorch Auto-Encoder Code (Deep Learning) durée heure minute seconde en bonne qualité , qui a été Téléchargé par l'utilisateur Alex Soupir 01 février 2024, Partagez le lien avec vos amis et connaissances, sur youtube cette vidéo a déjà été regardée 386 fois et il a aimé 7 téléspectateurs. Bon visionnage!
