machine learning from scratch

About ml algorithms.

classical machine learning

Linear Models

1. Linear regression analytical, gradient, sklearn
2. Support vector machine (SVM)
3. Logistic regression

Decisions trees

1. Decision Tree Regressor

Ensembles

1. Bagging on decision trees
2. Gradient Boosting

Probabilistic methods

1. Naive Bayesianс classifier scratch, sklearn
2. A little bit about statistical stability

Deep Learning

Neural Networks

Fully connected neural network

1. Task about two lines scratch, torch
2. Classifier handwritten numbers (MNIST) scratch, torch

Convolutional neural network

1. Classifier handwritten numbers (MNIST) torch

Natural language processing

1. word2vec word embeddings code, paper
2. LSTM next word prediction code, paper
3. LSTM text generation code, paper
4. seq2seq machine translation code, paper
5. seq2seq Attention machine translation code, paper
6. BERT code, paper

Computer Vision

1. Image clustering link, paper

materials:

ML YSDA

ML ITMO

NLP Lena Voita

ML MIPT

why?

how can you use algorithm without knowing how it works. Writing an algorithm helps me to understand all the nuances of its work, as well as in the next solution of the problem, you will see why it is good to apply one or another algorithm since you fully know their work

Tell me I didnt do it for nothing.

Omar Zoloev, 21. 26.06.2022