Links to papers I've read and gleaned insights
To Read:
NEURAL ARCHITECTURE SEARCH WITH REINFORCEMENT LEARNING
+
Learning What and Where to Draw
- Notes:
Recurrent Highway Networks
- Notes:
Bootstrapping Dialog Systems with Word Embeddings
- Notes:
Language Understanding for Text-based Games Using Deep Reinforcement Learning
- Notes:
Learning to Compose Neural Networks for Question Answering
- Notes:
Evaluating Prerequisite Qualities for Learning End-to-End Dialog Systems
- Notes:
Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation
- Notes:
DeepMath - Deep Sequence Models for Premise Selection
- Notes:
NLP Oriented
Building End-To-End Dialogue Systems Using Generative Hierarchical Neural Network Models
- Date: Jul 3, 2016
- Notes:
- Initialized their word embeddings with word2vec trained on google news dataset
- Used
End-to-end LSTM-based dialog control optimized with supervised and reinforcement learning
- Date: Jul 2, 2016
- Notes:
- 3 part system
- lstm - 32 cells, 1 layer deep
- parser
- state based software api thing
- 3 part system
Matching Networks for One Shot Learning
- Date: 6/15/16
- Notes: +
Learning End-to-End Goal-Oriented Dialog
- Date: 6/5/16
- Notes:
- storing conversational data (sec 4.)
Globally Normalized Transition-Based Neural Networks
- Date: 6/10/16
- notes:
- A.k.a. syntaxnet
- Feedforward network. Basis of improvement on normalization versus LSTM/residual networks.
- Original google blog post here.
Title: | Globally Normalized Transition-Based Neural Networks |
---|---|
date: | 6/10/16 |
- notes:
- a.k.a. syntaxnet.
- Feedforward network. Basis of improvement on normalization versus LSTM/residual networks.
- Original google blog post here.
Sequence to Sequence Learning with Neural Networks
Notes:
CV/CNN Oriented
RL Oriented
Deep Reinforcement Learning: Pong from Pixels
- Date: 6/15/16
- notes: +
Templates:
- Date:
- Notes:
Title: | title-of-paper |
---|---|
date: | date-read |
notes: | general notes |