Boost Fertility New Invention Ups Success Rates.pdf
Capturing the Mirage: Machine Learning in Media and Entertainment Industries
1. Machine Learning in Media and Entertainment
Industries
1
Sep 14, 2016
Zecca Lehn| Data Scientist
Domino Data Science PopUp LA 2016
Underlying Img: DKFindOut!
4. Failures might seem baffling if we follow our intuition and think of
artificial intelligence the same way we think about human intelligence.
Lukas Biewald What Can we Learn From AI’s Mistakes
4
By 2019, we believe that digital will account for more than
50 percent of the overall total for media.
McKinsey The State of Global Media Spending
Across media sectors, companies must ramp down traditional
offerings, ramp up new revenue streams, and innovate business
models.
BCG: Seeking Out Opportunities in a Dynamically Evolving Industry
5. WHERE_WE_ARE:
Sources: O’Reilly ‘The New Artificial Intelligence Market’ | Fidelity
5
IT:
• 32% AI Spending
• 31% SP500 MV
MEDIA & ENT:
• 2% AI Spending
• 5% SP500 MV
10. WHERE_WE_ARE:
Deep Natural Language Processing
(CNTK Now Open Source Github/MSFT)
10
Source: InformIT
O’Reilly:
• 3% of AI spending across all
industries...
11. Scene Completion | RNN LSTM| Plot Discovery
…getting closer to semantic continuity!
11
SUNSPRING Short| A Sci-Fi Starring Thomas Middleditch
13. WHAT_WE_SEE:
• DataFX integrates statistics
and special effects to create
new video and digital
content.
• Los Angeles based startup
with NBA and media clients
13
15. WHAT_WE_SEE:
15
•Presented at Big Data Day LA 2016
•“Prescriptive Analytics” applied
through client engagement in
feedback loop >> adding to lift
through levers
•Blending AI w/ econometric
models on BIG DATA
•Source: IRIS.tv / NBA
20. WHAT_WE_HEAR :
20
“Moneyball for Book Publishers” NY Times 2016
Sources: Publishing Perspectives, NPR
•Finish Book, Read Fast, Women/Man, Baby
Boomer?
•Startup from London
•Analytics to understand customers through
crowd engagement >> prediction >> channel
marketing & PR >> publishers
•Addresses 20/80 rule – 20% of books are
profitable, and 80% not… Disruptor?
•Free and sample ebooks offered in return for
your behavioral data
21. WHAT_WE_HEAR :
•Match directors and writers in pre-
production using NLP
•Using writers and directors past
projects, experience level, interests, etc.
•Takes descriptions of projects shared by
studios and production companies
•Aim to match equal ratio of men and
women directors
21
New LA based startup @DataPony
My name is Zecca Lehn, I’m an LA based Data Scientist, and I want to answer a question…
“Is AI and Machine Learning the New Frontier of Media and Entertainment?” I think so, but many see it as a mirage.
I’ll show why we need to change this mindset, and how it’s being realized by projects and companies in the industry’.
Let’s take a journey to find water… But how do we get there?
“What we’ll meet along the way”
WHERE WE ARE: Market conditions, and sentiment toward AI / ML
WHAT WE SEE: How it’s being integrated into visual content
WHAT WE HEAR: What companies and projects are delivering speech, music, and NLP
Looking forward…
What Can we Learn From AI’s Mistakes https://www.crowdflower.com/what-we-can-learn-from-ais-mistakes/
The state of global media spending http://www.mckinsey.com/industries/media-and-entertainment/our-insights/the-state-of-global-media-spending
Seeking Out Opportunities in a Dynamically Evolving Industry https://www.bcg.com/expertise/industries/media-entertainment/default.aspx
Industry breakdown: http://www.wikinvest.com/industry/Media_%26_Entertainment
Source: O’Reilly -- The New Artificial Intelligence Market 2016
Accenture The Promise of Artificial Intelligence : https://www.accenture.com/_acnmedia/PDF-19/AI_in_Management_Report.pdf#zoom=50
Accenture The Promise of Artificial Intelligence : https://www.accenture.com/_acnmedia/PDF-19/AI_in_Management_Report.pdf#zoom=50
Source Bloomberg: http://www.bloomberg.com/quicktake/artificial-intelligence
Note: NLP Accuracies are lower, but improving in the area of semantic prediction.
Practice: http://www.pyimagesearch.com/2016/08/10/imagenet-classification-with-python-and-keras/
Big Data Day LA ref:
http://www.slideshare.net/sawjd/big-data-day-la-2016-hadoop-spark-kafka-track-deep-learning-at-scale-alexander-kern-cofoundercto-pavlov
Source: http://www.informit.com/articles/article.aspx?p=2265404
MSFT DL Open Source CNTK: http://blogs.microsoft.com/next/2016/01/25/microsoft-releases-cntk-its-open-source-deep-learning-toolkit-on-github/
SUNSPRINT: https://thescene.com/watch/arstechnica/sunspring-sci-fi-short-film?utm_content=36474005&utm_medium=social&utm_source=twitter
Benjamin (automatic scripts) www.benjamin.wtf
Others:
ECLIPSE (Action/Adventure at Cannes w/IBM Watson): http://www.globalfuturist.org/2016/07/eclipse-the-worlds-first-ai-produced-short-film-hits-the-screens-at-cannes/
IMPOSSIBLE THINGS (Horror by Greenlight Essentials): http://www.ew.com/article/2016/07/26/artificial-intelligence-writes-perfect-movie-script
Recurrent Neural Networks RNN LSTM Github Example
http://karpathy.github.io/2015/05/21/rnn-effectiveness/
Mark Ruffalo Transformation into The Hulk
https://techcrunch.com/2016/07/11/walt-disney-co-reveals-9-new-startups-in-the-disney-accelerator-spanning-robotics-cinematic-vr-and-ai/
JELLYBOOKS : https://www.jellybooks.com/
CEO: http://publishingperspectives.com/2015/08/jellybooks-tracking-reader-engagement-for-better-marketing/#.V8soqpgrJhH
Publishers' Dilemma: Judge A Book By Its Data Or Trust The Editor's Gut?:
http://www.npr.org/sections/alltechconsidered/2016/08/02/488382297/publishers-dilemma-judge-a-book-by-its-data-or-trust-the-editors-gut