ml-core

This repository contains notes on different ml-based technologies. Please note that this is for my own educational purpose.

All the different post, should discuss the ideas in the following topics:

1. Idea - What is the main idea behind this approach?
2. Improvement - What does it make better than other approaches?
3. Concept - The concept in more details, and possibly pictures!
4. Examples - Own implementation of the technology
5. References - Links, Guides, Papers, etc.

Learning

Supervised Learning

1. TODO:

Unsupervised Learning

1. TODO:

Reinforcement Learning

1. TODO: [Q-Tables]
2. TODO: [Deep Q Networks]

Active Learning

1. TODO: [Active Learning]
2. TODO: [RL + AL]

Neural Networks

1. Linear Logistic Regression
2. TODO: [ANN/ MLP]

CNN - Convolutional Neural Networks

1. TODO: [CNN]

RNN - Recurrent Neural Networks

1. RNN
2. LSTM
3. GRU
4. TODO: [Sliced RNN]

Optimization

1. Gradient Descent
2. Greedy Search
3. BEAM Search

Encoder Decoder

1. Encoder Decoder

Attention

1. Attention

Transformer

1. TODO: [Transformer]
2. TODO: [Image Transformer]

AE - Autoencoder

1. TODO: [AE]
2. TODO: [VAE]

GAN - Generative Adversarial Networks

1. TODO: [GAN]

Metrics

1. Bleu
2. TODO: [Rogue]

Embeddings

1. Word Embeddings
2. TODO: [Spatial Embeddings]