search

yandex-research / rtdl-num-embeddings

(NeurIPS 2022) On Embeddings for Numerical Features in Tabular Deep Learning
https://arxiv.org/abs/2203.05556
MIT License
269 stars 31 forks source link

Sensemaking #17

Open firmai opened 1 month ago

firmai commented 1 month ago

How does this make senes, you basically just added another layer to the neural network an autoencoding layer. Is that what you mean?

Yura52 commented 4 days ago

Hi @firmai ! So sorry for the slow reply, is the issue still unresolved? Regarding your question, I think the TL;DR section of the readme can be helpful. Our idea indeed can be seen as adding a new special layer at the very beginning of a network, but the resulting model is still just a feed-forward network.