LOS ANGELES, NOVEMBER 7 2019
Applying AI & Machine Learning To Media & Entertainment
The data science salon is a unique vertical focused conference which brings together specialists face-to-face to educate each other, illuminate best practices and innovate new solutions in a casual atmosphere with food, great coffee and entertainment.
Get access to powerful decisionmakers in data science in an intimate setting at Data Science Salon LA, the only vertically-focused industry conference series around applications AI and Machine Learning in Media and Entertainment. Learn from practitioners, technical experts and executives how to solve real-world problems by harnessing disruptions in data, artificial intelligence, machine learning, and cutting-edge technologies. At DSS LA, we connect you with our powerful community face-to-face and digitally – each ticket comes with one year of access to DSS Insider, the content repository for all Data Science Salons.
Data Science Salons are one- or two- day events hosted at Blue Chip companies. Over 50% of our 200-500 attendees at each conference are data decisionmakers (Sr. Data Scientists and above). And we are the only data science conference with a gender balance in our speaking roster.
A Sample of Topics Covered
Enhancing the Viewer Experience with Machine Learning & AI
Content Personalization and Monetization
Personalization at Scale with AI
Cloud Automation and Machine Learning
Audience Targeting and Segmentation (across platforms)
Data and AI for emerging platforms
Meet Our Speakers
8.00AM – 8.55AM
8.55am - 9.00am
Jeremy Mason, Head of Operations at Formulatedby
9.00am - 9.30am
Literate Statistical Programming is not just about reproducibility
John Peach, Sr Data Scientist at Amazon Alexa
Science is facing a crisis around reproducibility and data science is not immune. Literate Statistical Programming is a workflow that binds the code used in an analysis to the interpretation of the results. While this creates reproducibility it also addresses issues around, auditing, re-usability and allows for rapid iteration and experimentation. This talk will describe a workflow that I have successfully used on small-scale data-sets in start-ups and on Amazon-scale problems in my work on Alexa. The talk will cover the tooling, workflow, and the philosophy you need to master Literate Statistical Programming.
About the speaker
A modern polymath, John possesses a unique and diverse set of skills, knowledge and experience. Having earned advanced degrees in Mechanical Engineering, Kinesiology and Data Science, his expertise focuses on machine learning, solutions to novel and ambiguous problems. He has a proven history of taking a problem from ideation to production by using a logical, but creative, data-driven approach. As a highly skilled Data Scientist, he has developed new techniques, lead teams, developed innovative data products and is a trusted advisor to decision makers.
John is a natural leader, customer focused, excellent communicator and problem-solvier. He loves new challenges and opportunities. His extensive background in software development, and modeling serve him well. His curiosity, creativity, focus and attention to detail have resulted in a track record of discovering hidden secrets in data.
John currently leads the Alexa Skill Store Science team. He works closely with engineering to build systems that will enable Alexa customers engage with third party applications, skills. The focus of his work is to help surface the right skill to the customer to achieve their goals. This includes arbitrating between skills, entity resolution, search, and personalization. He uses a data-driven approach to understand the customer’s intention and the capabilities of the skills in Alexa’s catalogue.
John fosters the growth of scientists by starting the Amazon Machine Learning University in Irvine plus the Alexa wide Data Science Excellence program. He frequently gives talks at universities and conferences. He is working to improve upon and formalize data science best practices. The focus has been on reproducible research. To that end he has developed an approach to improve data validation and reliability by using data unit tests. He has also developed the Data Science Design Thinking concept; to formalize and increase the efficiency of the analysis process.
9.30am - 10.00am
Hulu and Hang: Understanding Viewer Sessions with Big Data
Everyday, Hulu ingests 100 terabytes of user level app interaction data. This session data is the closest touchpoint we have to subscribers’ experience in our product short of joining them on their couch in their living rooms. Making meaning out of session data is a non-trivial effort across data instrumentation, engineering, analytics, and data science teams. In this talk, you will get an inside look at how we are tackling this monumental project at Hulu: from product design, to generating insights, to building predictive models – all to create the most personalized and engaging streaming experience for our subscribers.
Senior Data Analyst Lead at Hulu
Catherine has been at Hulu for the past 3 years. There, she stood up the Data Insights team, providing data analytics capabilities to support the operational and strategic decision-making of the business across Marketing, Product, Technology, Ad Sales, Content, and Finance.
Prior to Hulu, Catherine began her career in the operations research space with the Department of Defense supporting the acquisition and sustainment of billion dollar weapons systems. She then moved into the world of big data as a Data Analytics consultant for PwC focused on building out business intelligence environments for cable providers.
Catherine has a BA in mathematics from Providence College and an MS in Applied Mathematics from Northeastern University.
Data Scientist at Hulu
Herbie Huff is a data scientist at Hulu, where her projects include originals performance forecasting and Bayesian anomaly detection for performance of shows. Before coming to Hulu she worked for nearly a decade on data analytics and data science in the urban transportation field. She is a mathematician at heart, a recovering urban planner, and a lover of Los Angeles.
10.00am - 10.30am
How Spatial Analysis and Data Science Go Together
Shannon Kalisky, Product Manager – Analytics and Data Science at Esri
Location influences almost everything, from consumer behavior, to architecture, what we have access to, and the way we get around. It plays a part in who we reach, how we reach them, and can help us predict, understand, and quantify the impact that our decisions have on the human and natural world. This talk will cover how spatial thinking can help you build better models and tell stronger stories with your data.
About the speaker
Shannon Kalisky is the lead product manager for Analytics and Data Science at Esri, where she works with development and engineering teams to bring spatial data science mainstream. She started her career in GIS where she worked for a variety of organizations ranging from government to Fortune 500 companies, leveraging spatial data to uncover patterns and build predictive models with a combination of GIS and Python. Her undergraduate studies were in geography and her graduate education was in community and regional planning. She is currently pursuing her MBA in global business. When she’s not working behind a computer, you’re most likely to find Shannon with her hands dirty in a garden or at the local hardware store gathering supplies for her next project.
10.30am - 10.50am
Machine Learning for Contextual Targeting
Bruce Zhou, Senior Data Scientist at ZEFR
With the enrichment of video content on the Internet, the market of digital advertisement grows faster in recent years. However, due to the large scale and complexity of different platforms, more and more problems such as brand safety issues and insecure contents appear. Bruce is going to talk about how these problems attracts brands’ attention and how Zefr built a machine learning product to address these issues.
About the speaker
Bruce Zhou is a Senior Data Scientist at Zefr, and his daily responsibility includes machine learning and statistical modeling, model diagnostic, and development and productization of machine learning models. He has working experience in large-scale recommendation system and natural language processing. He graduates from the University of Washington at Seattle with a master degree in Statistics.
10.50am - 11.10am
Apply natural language processing in segmenting online audience
Jasmine Ngo, Manager, Analytics and Marketing Science at Deutsch
Sometimes, conversations on social media don’t reflect the mass’ sentiments accurately (think about people who rarely use Twitter or have a public profile on Facebook). That’s when local news come into play – local articles can sometimes reflect sentiments on certain topics by specific areas. By scraping thousands of articles online and using NLP / other methodologies to analyze them, we can get interesting insights on different topic. This talk introduces a tool to do that.
About the speaker
Jasmine’s varied experiences include leveraging social media campaigns to research audience insights and analyzing multi-source big data to refine marketing strategy for big companies. Her analytics capabilities have been sharpened through work at Experian and ad agency Deutsch. At Deutsch, she has leveraged in-depth data analytics and customer insights to inform successful marketing strategies for Taco Bell, 7-Eleven, Zillow, CMAB and H&R Block. Prior to getting her MBA in data analytics at UC Irvine, she was a successful TV producer with national audience choice award and multiple high-rating shows in Vietnam.
11.10AM – 11.40AM
Coffee Break & Networking
11:40 am - 12:05 pm
Machine Learning & AI: Demystified for TV Advertising
Diane Yu, CTO and Cofounder at FreeWheel, Comcast
About the speaker
Diane Yu is the chief technology officer for Comcast Advertising as well as the co-founder of FreeWheel, A Comcast Company. In this capacity, Diane leads the architecture and operation of FreeWheel’s video monetization platform, with oversight of a global engineering team spanning the Americas, Europe and China. Diane also oversees the engineering operations and complete lifecycle product development, and leads organizational and high level technical architecture design.
Over the past 10 years, she built and evolved the company’s research and development hub in her native Beijing. Today, the organization is comprised of over 600 engineers and ops employees in Beijing, France, Philadelphia, Chicago and New York — 30 percent of whom are female engineers, reflecting Diane’s commitment to diversity and inclusion in the workplace.
12:05 pm - 12:25 pm
Measure Ad Effectiveness
Sagnik Mazumder, Data Scientist at Samba TV
To understand the effect of an advertisement, conversion rate alone does not answer the question of impact. We want to measure the incremental effect an advertisement has on the exposed audience, accounting for the likelihood that some portion of the exposed audience would have converted regardless of exposure. In digital advertising, this effect is quantified by randomly selecting some of the people that would have been exposed to the digital ad and instead serving them a “placebo ad,” placing them into a control group. In general, most broadcast TV commercials are an all-or-nothing scenario: everyone watching the program gets the same commercial. When this exposed group is compared to all unexposed households, only limited conclusions can be drawn, because they are fundamentally different groups of households with different conversion propensities. The unexposed group will contain a higher proportion of households that are not in the targeting strategy and that might have a very low likelihood of conversion. The difference between the exposed group’s and the unexposed group’s conversion rates thus gives an artificially high incremental lift. Samba TV’s proprietary Synthetic Control Group (SCG) process allows us to overcome this hurdle by deriving a control group using a household’s TV viewership data. We design the control group to be, based on our data, the target group that was most likely to see a specific ad campaign spot but that did not. Comparing the exposed group’s conversion rate to the conversion rate of this control group then reveals a measurable, attributable lift.
About the speaker
Sagnik joined Samba TV in 2016 and has since been a crucial member of the Data Science and Analytics team; he currently works in the role of Data Scientist. His initial responsibility was to build attribution models to measure the impact of ads and compare how TV ads and digital ads played a role in conversions. Sagnik has also been a lead developer of Samba TV’s first-of-its-kind Synthetic Control Group methodology, which has received extremely positive feedback from multiple clients across a variety of industry verticals.
12:25 pm - 12:45 pm
How Graph Technology is Changing AI
Jake Graham, Lead Product Manager – AI and Graph Analytics at Neo4J
Graph enhancements to Artificial Intelligence and Machine Learning are changing the landscape of intelligent applications. Beyond improving accuracy and modeling speed, graph technologies make building AI solutions more accessible. Join us to hear about 4 areas at the forefront of graph enhanced AI and ML, and find out which techniques are commonly used today and which hold the potential for disrupting industries. We’ll provide examples and specifically look how: – Graphs provide better accuracy through connected feature extraction – Graphs provide better performance through contextual model optimization – Graphs provide context through knowledge graphs – Graphs add explainability to neural networks
About the speaker
Jake Graham leads product management for Neo4j’s Graph AI and Analytics team. With several years experience developing and implementing graph approaches to data science, Jake has earned valuable insight into how to use graphs to understand network structure in enterprise settings. Prior to joining Neo4j, Jake was Director of Product Management for Intel’s Saffron technology, which also focused on leveraging graph approaches to AI.
12.45PM – 1.45PM
Lunch Break & Networking
1:45 pm – 2:05 pm
Being Agile when Developing AI Products
Manasi Vartak, Co-founder and CEO at Verta.AI
Models are the new code. While AI is increasingly driving key products and business processes, we find that robust tools and development practices are missing for AI & ML, leading to the slow development of AI products, brittle AI models, and long times to market. As the team behind Verta.AI and ModelDB from MIT, we have helped develop, manage, and deploy hundreds of models ranging from cutting-edge deep learning models to traditional ML models in finance. We have found one theme that consistently differentiates successful AI product development — these product teams have all adopted an “Agile” methodology including extensive tooling support, agile processes, and team structures. In this talk, I will distill the best practices we have learned in being agile while developing AI products and describe how organizations can implement agile to enable rapid development and delivery of AI.
About the speaker
Manasi Vartak is the founder and CEO of Verta.ai (www.verta.ai), an MIT-spinoff building software to enable high-velocity machine learning. The Verta platform enables data scientists and ML engineers to robustly version ML models, collaborate and share ML knowledge, and when models are ready for graduation, to deploy and monitor models in production environments. Verta grew out of Manasi’s Ph.D. work at MIT on ModelDB, the first open-source model management system deployed at Fortune-500 companies. Manasi previously worked on deep learning for content recommendation as part of the feed-ranking team at Twitter and dynamic ad-targeting at Google. Manasi is passionate about building intuitive data tools, helping companies become AI-first, and figuring out how data scientists and the organizations they support can be more effective. Manasi has spoken at several top research as well as industrial conferences such as SIGMOD, VLDB, SparkSummit, and AnacondaCon, and has authored a course on model management.
2:05 pm – 2:35 pm
Sylvia Tran, Data Scientist at Gracenote
User preferences and content similarity are both key to recommendation systems. While content similarity has been widely explored and utilized by many companies in the media & entertainment industries, it still remains relevant as the amount of data and metadata available continues to grow and change. This talk discusses some of the challenges of content similarity and explores a few different attribute groups (aside from genre and cast) by which content similarity can be measured.
Traditional attributes, like genre and cast alone, may not be as additive as they once were. More specifically, movies like Ted (starring Mark Wahlberg) and Shaun of the Dead do not neatly fit into a single genre.
This talk also demonstrates how certain tried and true similarity metrics still yield meaningful and reasonably interpretable results for media & entertainment.
About the speaker
Sylvia is a Data Scientist at Gracenote where she is responsible for working closely with both product and engineering to deliver data science solutions for the development of new products, or to enhance existing products that serve businesses in the media & entertainment industry.
Prior to Gracenote, Sylvia was a data scientist at OpenDrives, and taught data science at General Assembly. Prior to her experience in data science, she was a Director & VP at Wells Fargo Capital Finance LLC working in the capital markets primarily on mergers & acquisitions.
Sylvia graduated from UCLA with a BS in Applied Mathematics. She is currently completing her MS from Georgia Tech in Analytics-Computational Data. Sylvia currently serves as the chairperson for the Los Angeles chapter of PyLadies (a Python Software Foundation backed organization that promotes women’s more active participation in the Python open-source community).
2:35 pm – 3:05 pm
How Graph Technology is Changing AI
Shilpi Bhattacharyya, Data Scientist at IBM
Who does not love the American television sitcom – Friends? And we definitely want to learn what makes this sitcom so popular. Can the most important aspects of some of the top shows of all the times be related? Is there something common which makes them a success? If not, can we find out and draw a correlation amongst them? In this talk, I would demonstrate the essential elements of few of these most successful sitcoms which have helped them connect with the audience at such a massive scale around the world. I would use data science and machine learning techniques as sentiment analysis, data visualization and correlation graphs on the transcripts available for these sitcoms to achieve the results. I would also focus briefly on the favorite characters. I believe this work would be able to bring out a concrete answer to the apparent question amongst the makers to understand the reasons which makes a hit show, with evidence backed up by data science.
About the speaker
Shilpi Bhattacharyya is a Data Scientist in the Cloud Garage Solution Engineering team at IBM. She graduated from Stony Brook University with her research thesis on Big Data computing under the supervision of Dr. Dimitrios Katramatos and Dr. Barbara Chapman. Previously, she was a guest researcher at Brookhaven National Laboratory, where she collaborated with other computer scientists to solve emerging data science problems and wrote papers about her findings. In the past, she has also worked with technology companies as Oracle, Samsung Research Institute and Sapient Corporation in various engineering roles. She has published in Elsevier journal and has spoken at multiple international conferences about her work. She has a vision for a data literate world.
3:05 pm – 3:35 pm
Coffee Break & Networking
3:35 pm – 4:05 pm
Ad Planning & Operations: Predictive Analytics for Resource Allocation
Leondra James, Manager, Analytics and Operations at Saatchi & Saatchi
Saatchi & Saatchi is global, full service advertising agency / creative communications network. Learn about the interesting questions they’re asking and how they’re leveraging data to answer them. Entails broad overview of predicting advertising initiative resources and their respective allocations.
About the speaker
Leondra is a Columbus, OH native, Los Angeles based data scientist with a rich background in entertainment and media analytics. Today, she manages Saatchi & Saatchi’s operations predictive analytics unit for Toyota, where she oversees the implementation of statistical & predictive modeling of human resource & budget allocation to advertising campaigns. She works closely with project scoping personnel and senior management to improve business processes, project labeling, similarities and resource projections.
She also has a rich, educational background – a current PhD Candidate for Information Technology, she studied Music and Business with a focus in Business Management at Otterbein University where she earned a scholarship for violin performance. Additionally, she has a Master in Entertainment Industry Management where she focused in Business Analytics from Carnegie Mellon University Heinz College of Information Systems Management and Public Policy, an MBA with a focus in Statistics, Data & Decisions from Smartly Institute, and a post graduate professional certificate in Data Science from Harvard University.
Beyond work, Leondra enjoys playing video games and table-top / RPG games with her friends, metal shows and unrealistic action films.
4:05 pm – 4:35 pm
On the Road to More Holistic Player Understanding
Wesley Kerr, Principal Data Scientist at Riot Games
One of the challenges seemingly all data scientists face is finding a clean data set which contains the state of the player right before they take some event in the system. Typically we need to interact with event-driven systems and/or databases that only maintain the current snapshot of the player. In this talk we highlight some work we have done to recreate the up to date snapshot of the player captured before each event and demonstrate how we can leverage this dataset to improve personalization and model the players’ likelihood to churn.
About the speaker
Wesley is a Principal Data Scientist at Riot Games and leads a research team investigating how AI can be leveraged to improve the game design and the player experience. He helped launch League of Legends’ “Your Shop” multiple times for our players, has focused on detecting unsportsmanlike behavior in League of Legends, and more recently has developed models predicting the likelihood of a player churning from League. Prior to joining Riot he spent time in Google Research and working on search and recommendations for Google Play. He received his PhD in Computer Science from the University of Arizona.
4:40 pm – 5:20 pm
Amarita Natt, Managing Director, Data Science at Econ One Research
Dr. Amar Natt is an economist and data scientist specializing in econometric modeling of large-scale (multi-terabyte) transactional datasets. Her recent work has included predictive modeling of major airlines’ loyalty programs for the purposes of booking financial statements. This work includes customer profiling, predictive modeling given current conditions, and scenario modeling incorporating potential changes to programs or industry conditions. In the litigation space, Dr. Natt assists in model design and data analysis for price-fixing and antitrust litigation cases. Outside of EconOne, Dr. Natt is a lecturer at UCLA, teaching a Data Science seminar of her own design.
Noelle Saldana, Director of Product Management, Data Science & Analytics at Salesforce
Noelle is currently the Director of Product Analytics and Data Science at Heroku (Salesforce), where she is driving how Heroku captures, analyzes, and leverages data throughout the organization including Product, Marketing, and Design. She was previously a leader for Pivotal’s Data Science team, where she worked with numerous enterprise and startup companies planning and executing their analytics initiatives.
She is passionate about the power of Math and making it more accessible; she enjoys collaborating with other disciplines and being creative about using analytics, doing pro bono Data Science work through DataKind, and doing her part to encourage students in underrepresented groups to pursue STEM careers.
Noelle holds an A.B. From Washington University in St. Louis in Applied Mathematics and Physical Anthropology and a M.S. in Applied Mathematics from Cal Poly Pomona.
Natasha Ericta, Head Of Analytics & Data Science at Jukin Media
Natasha has a background in Applied Mathematics, Programming and Statistics from UCLA. As a statistician by training, she applies statistical modeling, programming and theory to a variety of use cases and real world scenarios. She brings 15+ years of analytical experience and data science to business applications, consulting and strategy including vendor and data science management and scaling across multiple industries and verticals. As a Westside Angeleno, Natasha also enjoys community involvement in neighborhood council activities, animal rescues and local arts and entertainment.
Kate Coke, Executive, Data Insights & Analytics at CAA
Kate Coke leads the Data Insights team for CAA, a leading entertainment and sports agency representing many of the most successful and influential professionals working in Film, Television, Music, Sports, Theater, Digital Content, Video Games, and more.
Having joined CAA in 2014, Coke brings more than 13 years of entertainment research and insights experience with her, drawing from her previous tenure at FX Networks, Warner Bros. Television, CBS Television, and 20th Television. She is adept at leveraging state of the art toolsets to deliver actionable findings that inform business decisions and unlock value for clients. She is passionate about creative storytelling through data, uncovering new market trends, education, and the intersection of entertainment and technology.
5:20 pm – 5:50 pm
Utilizing Data and Data Science to Optimize Tune-In
Diane Leung, Principal, Analytics Innovation at Altman Vilandrie & Company
Measuring and optimizing tune-in is critically important for the media and entertainment industries. This discussion focuses on best practices for utilizing data and machine learning to optimize tune-in on national linear inventory. In order to do this properly, advertisers need to unify their marketing ecosystem, design a holistic measurement approach, and break down barriers for closed-loop, incremental measurement. In this session you will learn how to: 1) Create a framework for utilizing data and machine learning to maximize tune-in and 2) Overcome analytical obstacles created from fragmented and incomplete data.
About the speaker
Diane Leung is a Principal in Altman Vilandrie & Company’s Boston headquarters and leads the firm’s Analytics Innovation Team (AIT), which provides a range of cutting-edge analytic and data services for the firm’s Telecommunications, Media, and Technology (TMT) sector and private equity clients. Under Diane’s direction, AIT provides Altman Vilandrie & Company’s clients with access to the full range of data and analytics services, including Big Data Analysis and Engineering, AI/Machine Learning, Geospatial Analytics and Data Visualization. Examples of client projects supported by AIT include marketing and sales optimization, pricing strategy, competitor analysis, infrastructure deployment, and program efficiency.
Diane is an expert in media analytics and attribution with a focus on advanced advertising for TV. Her TV clients have relied on her for her expertise in working with STB data, omnichannel audience targeting and segmentation, large-scale modeling and scoring, and linear and addressable campaign analytics (including test & control design, closed loop attribution, media schedule optimization, and more). She is an expert in designing analytic methodologies to overcome data coverage gaps and reporting biases.
Diane has over 14 years of experience, most recently as Director, Data Science & Analytics at Acxiom, a global marketing services and technology company. She also held multiple positions at the database marketing firm The Allant Group and at the digital marketing firm Digitas after five years in economic consulting. Diane earned a Master’s degree in Marketing Science from Columbia Business School, and a B.A. in Economics from Wellesley College.
6.00PM – 8.00PM
And many more!
DATE AND TIME
November 7, 2019
8:00 AM – 8:00 PM PDT
The East Angel
670 S Anderson St
Los Angeles, CA 90023
t: (415) 322-0760