Jill-Jênn Vie
Researcher at Inria
% 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}} —
Introduction
Outline
Mangaki
- Recommender system for anime & manga
- Why nonprofit?
Research
- The geometry of mangas
- Deep learning: learning from massive data (it’s all blocks)
- Our new algorithm (name will come, wait for it)
Future
- ???
Mangaki, recommendations of anime/manga
Rate anime/manga and receive recommendations
2,000 users, 10,000 anime/manga, 350,000 ratings
- myAnimeList (RIP their API)
- AniDB
- AniList
- (soon) TVtropes
Mangaki
Mangaki
Build a profile
Mangaki prioritizes your watchlist
Browse the rankings: top works
Why nonprofit?
- Why should blockbusters get all the fun/clicks/money?
- Maybe there is one precious, unknown anime \alert{for you}
- and we can help you find it
- Driven by passion, not money
Everything is open source: github.com/mangaki (Python, Vue.js)
Awards: Microsoft Prize (2014) Japan Foundation (2016)
Browse the rankings: precious pearls
Research
Embedding you in the geometry of mangas
\vspace{-3mm}
Many blocks for recommendation
\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)
A new block appears!
Illustration2Vec (by Masaki Saito & Yusuke Matsui, 2015)
\centering
{width=42%}
{width=45%}
We have (many) posters
Existing algorithms
\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}
Our new algorithm
Combine blocks
Use machine learning to select the best model
\hfill Blended Alternating Least Squares with Explanation
Our new algorithm
Combine blocks
Use machine learning to select the best model
\hfill Blended Alternating Least Squares with Explanation
Our new algorithm
Combine blocks
Use machine learning to select the best model
\hfill Blended Alternating Least Squares with Explanation
Our new algorithm
Combine blocks
Use machine learning to select the best model
\hfill Blended Alternating Least Squares with Explanation
Our new algorithm
Combine blocks
Use machine learning to select the best model
\hfill Blended Alternating Least Squares with Explanation
Our new algorithm
Combine blocks
Use machine learning to select the best model
\hfill Blended Alternating Least Squares with Explanation
BALSE
\centering
Our paper BALSE was accepted at the MANPU workshop!
\pause
Genuine excerpt from the reviewers
\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.
Future
Make your neural network look at the manga
Extract frames from episodes
\hfill \emph{Cowboy Bebop EP 23} “Brain Scratch”, Sunrise
Watching assistant
Take home message
- Using machine learning, you can \alert{predict the future}
- Machine learning is made of many blocks put together
- But they need data
- To get data, interoperability is super important
- Let’s work together to pair myAnimeList / AniDB / TVtropes!
- \alert{Contact us} to know more
Still, there are things we can never predict
\centering
{width=50%}
Thanks! Do you have any questions? \hfill jj@mangaki.fr
mangaki.fr/map \hfill github.com/mangaki