% Using Posters and Deep Learning\newline to Recommend Anime & Mangas % Jill-Jênn Vie, PhD\newline RIKEN Center for AIP, Tokyo\vspace{3mm}\newline \includegraphics[width=2cm]{figures/aip-logo.png} % July 5, 2018 — handout: true section-titles: false header-includes: - \usebackgroundtemplate{\includegraphics[width=\paperwidth,height=\paperheight]{figures/bg.png}} —
Rate anime/manga and receive recommendations
2,000 users, 10,000 anime/manga, 350,000 ratings
Everything is open source: github.com/mangaki
(Python, Vue.js)
Awards: Microsoft Prize (2014) Japan Foundation (2016)
\vspace{-3mm}
\begin{itemize} \item[KNN] $K$-Nearest Neighbors \item[ALS] Alternating Least Squares \item[FMA] Factorization Machines \end{itemize}
Input: ratings
Output: representation of users and items
(geometry used for reordering & recommendation)
Illustration2Vec (by Masaki Saito & Yusuke Matsui, 2015)
\centering
{width=42%} {width=45%}
\begin{itemize}
\item[ALS] Take ratings $\rightarrow$ recommendations
\emph{Does not work when no ratings are available (ex. new work)}
\item[LASSO] Take tags/ratings $\rightarrow$ recommendations
\emph{Does not take care of people with similar taste}
\item[I2V] Take posters $\rightarrow$ tags
\end{itemize}
Combine blocks
Use machine learning to select the best model
\hfill Blended Alternating Least Squares with Explanation
Combine blocks
Use machine learning to select the best model
\hfill Blended Alternating Least Squares with Explanation
Combine blocks
Use machine learning to select the best model
\hfill Blended Alternating Least Squares with Explanation
Combine blocks
Use machine learning to select the best model
\hfill Blended Alternating Least Squares with Explanation
Combine blocks
Use machine learning to select the best model
\hfill Blended Alternating Least Squares with Explanation
Combine blocks
Use machine learning to select the best model
\hfill Blended Alternating Least Squares with Explanation
\centering
\pause
\vspace{3mm}
Two models individually are existing methods, but this work presents a novel fusion method called \alert{Steins gate} to integrate results given by two models.
Extract frames from episodes
\hfill \emph{Cowboy Bebop EP 23} “Brain Scratch”, Sunrise
\centering
{width=50%}
mangaki.fr/map
\hfill github.com/mangaki