Recurrent Neural Networks with Attention and why they are so hard to train

Hosted By
Lev K.
Details
Dear Data Science Journal readers!
Please read the materials before the meetup. If you don't get a chance, feel free to come join the discussion anyway
-
Motivation: "Teaching Machines to Read and Comprehend" (http://arxiv.org/abs/1506.03340) by Deepmind.
-
A math and code treatment of attention in NN - looking for advice in the comments.
One idea: Recurrent Models of Visual Attention (http://gitxiv.com/posts/ZEobCXSh23DE8a8mo/recurrent-models-of-visual-attention)
- Pascanu "On the difficulty of training Recurrent Neural Networks (http://www.jmlr.org/proceedings/papers/v28/pascanu13.pdf)"
Introductory blog posts, theses and ipython notebooks are very welcome in the comments.
Message to those on the waiting list: 20 more places will be released 2 days before the event.
London Data Science Journal Club
Moss House, 15-16 Brook's Mews, London W1K 4DS · London
Recurrent Neural Networks with Attention and why they are so hard to train
0 spots left