https://gilgi.org/enContents © 2020 <a href="mailto:site@gilgi.org">gilgi</a> Sat, 23 May 2020 20:33:04 GMTNikola (getnikola.com)http://blogs.law.harvard.edu/tech/rsshttps://gilgi.org/blog/dota-hero-embedding/gilgi<div class="cell border-box-sizing text_cell rendered"><div class="prompt input_prompt"> </div><div class="inner_cell"> <div class="text_cell_render border-box-sizing rendered_html"> <p><img src="https://gilgi.org/images/blog/dota-hero-embedding.png" alt=""></p> <p>One of the coolest results in natural language processing is the success of word embedding models like <a href="https://en.wikipedia.org/wiki/Word2vec">Word2vec</a>. These models are able to extract rich semantic information from words using surprisingly simple models like <a href="https://en.wikipedia.org/wiki/Bag-of-words_model#CBOW">CBOW</a> or <a href="https://en.wikipedia.org/wiki/N-gram#Skip-gram">skip-gram</a>. What if we could use these generic modelling strategies to learn embeddings for something completely different - say, Dota 2 heroes.</p> <p>In this post, we'll use the <a href="https://docs.opendota.com/">OpenDota API</a> to collect data from professional Dota 2 matches and use <a href="https://keras.io/">Keras</a> to train a Word2vec-like model for hero embeddings.</p> <p><a href="https://colab.research.google.com/github/gilgi/gilgi.github.com/blob/src/posts/dota_hero_embedding.ipynb"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"></a></p> <p><a href="https://gilgi.org/blog/dota-hero-embedding/">Read more…</a> (17 min remaining to read)</p></div></div></div>Dota 2embeddingkerasmachine learningnotebookPythonWord2vechttps://gilgi.org/blog/dota-hero-embedding/Sun, 27 May 2018 04:00:00 GMT