Learning Sequential Data
Overview
Human Lives are but un-solved sequences, but of-course we want AI to learn them !
Outline
Previous Talks
- Spring 2019 : Sequences and Language Models
- Fall 2019 : Graphs and Stuff
Resources
Attention and Language Models
Papers
- Transformers : Attention is all you need
- The Annotated Transformer
- Pretraining Language Models
- BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Blogs n Tutorials
- Understanding Attention Mechanism
- Lil’s Log: Attention
- Pytorch Transformer Tutorial
- Attention Mechanism n Memory Networks : skymind.ai
Vids
- Attention Model : deeplearning.ai
- Attention and Memory in Deep Learning : deepMind google
- Attention Paper Review
- Bert Explained
- Bert:TDLS
Graphs
Papers
- Graphs Survey 1
- Graphs Survey 2
- Graphs Survey 3
- Representation Learning on Graphs: Methods and Applications
- More Graph papers
Blogs n Tutorials
- Graph Convolution Neural Network
- Stanford Network Representations
- What is network representation learning and why is it important?
- Learning low-dimensional embeddings of nodes in complex networks (e.g., DeepWalk and node2vec).
- Techniques for deep learning on network/graph structed data (e.g., graph convolutional networks and GraphSAGE).
- Applications of network representation learning for recommender systems and computational biology.
Vids
Recurrent Neural Nets
- Original Recurrent Net
- RNN in Numpy and Passage Generation
- Pytorch RNN Beginner 1
- Pytorch RNN Beginner 2
Long Short Term Memory
- Original Lstm Tutorial - Jurgen Schmidhuber
- Understanding Lstms
- Lstm in Numpy from Scratch
- Pytorch Lstm Beginner
- Pytorch Lstm Seq2Seq
Language Models
Word Vectors
- Word2vec - Original Paper
- Training Millions of Embedding Model
- What are word embeddings
- Understanding word2vec
- Word Embedding Applications
- Word2vec using Gensim - Tutorial
- How to use Pre-Trained Word2vec Embeddings in Pytorch
- ELMO - Deep contextualized word representations
- COVE; Learned in Translation: Contextualized Word Vectors