Algoliterary Encounters: Difference between revisions
From Algolit
(→text-to-numbers / vectors) |
|||
Line 43: | Line 43: | ||
* [[word embeddings]] one-hot-vector script | * [[word embeddings]] one-hot-vector script | ||
− | * [[word2vec_basic.py]] - annotated word2vec script | + | * [[word2vec_basic.py]] - annotated word2vec script |
===== layers, nodes, weights ===== | ===== layers, nodes, weights ===== |
Revision as of 08:22, 4 October 2017
Start of the Algoliterary Encounters catalog.
General Introduction
Overview of Techniques
Rule-based agents
Classic Machine Learning
Neural Networks
Elements of Neural Networks
input (datasets)
- posters with literary works that are readable for machines and escape copyright
- public domain dataset
text-to-numbers / vectors
- posters of matrices
- bag-of-words: physical book vector exercise (with multiple books or with one book)
- word embeddings one-hot-vector script
- word2vec_basic.py - annotated word2vec script
layers, nodes, weights
- backpropagation (linear algebra, influence of weight on network)
multidimensionality
- digital interactive visualisation & printed visualisation 1 poster for 1 dimension (total could be 30 posters)
algorithms/nodes
- softmax poster/booklet with comments on code & output
- visualisations
- playground.tensorflow.org javascript
output
Datasets
- Context of Neural Networks
- Frameworks & Existing Communities
- Algoliterary Works by Others