shayneobrien/ml-from-scratch

Stars
1
Language
Jupyter Notebook
Created over 5 years ago
Updated over 5 years ago

shayneobrien/ml-from-scratch

shayneobrien

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Implementations of basic machine learning algorithms using only numpy, pandas, and matplotlib.

generative-models

Annotated, understandable, and visually interpretable PyTorch implementations of: VAE, BIRVAE, NSGAN, MMGAN, WGAN, WGANGP, LSGAN, DRAGAN, BEGAN, RaGAN, InfoGAN, fGAN, FisherGAN

Jupyter Notebook

coreference-resolution

Efficient and clean PyTorch reimplementation of "End-to-end Neural Coreference Resolution" (Lee et al., EMNLP 2017).

numerical-methods

Methods in numerical analysis. Includes: Lagrange interpolation, Chebyshev polynomials for optimal node spacing, iterative techniques to solve linear systems (Gauss-Seidel, Jacobi, SOR), SVD, PCA, and more.

language-modeling

Language modeling on the Penn Treebank (PTB) corpus using a trigram model with linear interpolation, a neural probabilistic language model, and a regularized LSTM.

Jupyter Notebook

sentiment-classification

Neural sentiment classification of text using the Stanford Sentiment Treebank (SST-2) movie reviews dataset, logistic regression, naive bayes, continuous bag of words, and multiple CNN variants.

Jupyter Notebook

explicit-gan-eval

Code for reproducing the results of "Evaluating Generative Adversarial Networks on Explicitly Parameterized Distributions" (O'Brien et al., NeurIPS 2018).

text-segmentation

Neural and nonneural text segmentation methods.

Jupyter Notebook

text-cluster

Offline and online (i.e., real-time) annotated clustering methods for text data.

Jupyter Notebook

machine-translation

Neural machine translation on the IWSLT-2016 dataset of Ted talks translated between German and English using sequence-to-sequence models with/without attention and beam search.

Jupyter Notebook

selected-papers-and-presentations

A sample of selected papers that I have authored or co-authored.

conversational-analysis

Unsupervised methods for analysis of conversational transcripts

lexrank

LexRank for ranking documents containing some keyword or keyphrase using cosine similarities of either naive, tfidf, or idf-modified-cosine. Non-query ranking also supported.