Machine Learning in production: From research to the customer
If you had to describe the Machine Learning process in 5 steps, from research to customer, what would they be? This is a question I often ask candidates interviewing for the Zappos Data Science team. My hope is that one of them will be able to tell me, so I can stop trying to figure it out myself. There used to be a time when we didn’t have enough data but we had ideas. Today we not only have the data, but we also have the ideas implemented as models. The real challenge now is to put those models into production. In today’s talk we will go over the Dos and Donts of deploying various Machine Learning and Artificial Intelligence methodologies at scale and bridging the gap between research and production environments.
Democratizing Data Science Within Your Company
You wouldn’t expect only professional writers to know how to write. So why would you expect only professional data scientists to know how to analyze data? Advancements in Python and R education have made programming for data science more intuitive than ever. With the right training, everyone from product managers to financial analysts to marketers can bring insights to their team.
Data-Driven Decision Making
Effectively aggregating & delivering real information to drive educated decision making.
How I Learned to Quit Worrying and Trust the Crowd
Several years ago, Topcoder solved a complex International Space Station energy production challenge… but had to first learn a hard lesson—a lesson that became a guiding principle for crowdsourcing as we know it. Join Andy LaMora, Global Director for Crowd Analytics & AI at Topcoder, and learn firsthand how Topcoder’s ISS work springboarded crowdsourcing into the future. He’ll also detail how Fortune 500 companies and government agencies alike are leveraging a crowd-powered consulting model to innovate, solve problems on a global scale, and develop unique business solutions.
Networking Coffee Break
Microtransit: Innovation as a Key Driver
Microtransit is a current topic of discussion in the IoT and Smart Cities
world. Innovation is key to drive change
when it comes to transportation planning and execution. This panel of experts in logistics,
operations, robotics and manufacturing bring together a model that will
revolutionize the way people get around in their cities. By partnering, these companies have developed
an innovative and scalable solution utilizing neighborhood electric vehicles to
connect people to public transportation as well as getting accurate positioning
systems in place (Smart Roads) and real-time data through telematics
devices. By adding such a service,
communities can change their attitudes about using their cars on a daily basis,
reducing carbon emissions, traffic congestion and driving economic
re-development in transit or food deserts.
Deploying Data Science: A Blueprint for Success
Deploying data science projects is difficult for any organization. For organizations looking to compete with the disruptive forces of GAFA-like companies (Google, Amazon, Facebook, Apple) the successful deployment of data science projects is imperative to their long-term success. In this talk we will discuss people and process strategies to quickly prototype and deploy data science at scale. We will discuss several strategies of data science team building and several alternative methods of deployment. The talk will close with a detailed description how a current customer builds and deploys product recommendations as a REST API.
Twitch Insights: Broadcaster, Viewer, and Player Research and Integration
Kristin’s talk will focus on Twitch Insights: Broadcaster, Viewer,
and Player Research and Integration. We’ll explore the current state and future
of Twitch Insights--self-serve data, analytics, and actionable
recommendations--as well as how Twitch Integrations tie into the
A Hybrid Approach to “Boosting” Machine Learning Models
For machine learning
and AI to reach its full potential, other elements of supervised learning are
often required to “boost” model performance and optimize results. Boosting is a
term where weak models are made into strong models. Adaptive Boosting is
a machine learning meta-algorithm that can be used in conjunction with
many other types of learning algorithms to improve performance. In
practice, it can be used in a variety of industries and applications. This
session will focus on hybrid or ensemble approaches of machine learning that
combine the inherent value of unsupervised attributes with specific techniques
of Supervised Learning, Reinforcement Learning and Computer Vision. Together,
this combination outperforms stand-alone applications of machine learning or
AI, delivering greater autonomy, speed, accuracy, and agility.
Many data science practitioners find themselves confronted
with free text data. Often, the goal is simply to extract features from the
text that are useful for predictive modeling, rather than performing a full
semantic analysis. Latent Dirichlet Allocation is an unsupervised learning
method that extracts groups of related words, or "topics", from a
collection of documents. Surprisingly, the theme of each topic is usually
obvious to a human reader, indicating that this method has uncovered real information
about the documents.
Using a collection of 50,000 movie reviews, we extract
topics from the review text, then use these topics to create a model that
predicts the review sentiment with very high accuracy. We will also cover the
conceptual underpinnings of LDA as originally introduced by Blei et. al.
Success Stories with Prescriptive Analytics in the Gaming Industry
The world of gaming is moving at a rapid pace with a wide variety of innovation in traditional casino games, skill-based games and sports betting, combined with an ever-evolving player demographics. Casino operators are constantly looking to understand their players better to provide an exciting, personalized gaming experience and establish brand loyalty. Scientific Games is the market leader in the Gaming and Lottery space, with casinos around the world utilizing Scientific Games’ software solutions to improve their operations. Leveraging the power of Big Data, Data Science and Prescriptive Analytics, Scientific Games has helped casinos transform to an analytics-driven organization. In this presentation, we look at how the journey of Advanced Data Analytics started at Scientific Games, the challenges and successes, the road ahead and what it means to create an analytics culture in your organization.
Networking Coffee Break
Lessons learned in shifting a legacy culture towards data, testing, and personalization
Over the last couple of years, Tribune Interactive has made a major shift from being a company of analysts working with vendor tools and Excel to adopting more modern big data best practices. In this talk, we'll cover some of the learnings we've had around driving adoption of data usage and tools in teams that have not been particularly data-driven in the past, moving from a tabular to designed data view in dashboarding and data presentation, and challenges in getting stakeholders to understand what a model is and how to use it.
Excuse me, are we data driven?
With today’s abundance of Big Data technologies we all would like to think of our organizations as being data driven. But is that really the case? Investing lot in a data platform does not immediately make a company data driven for the very same reason that buying an expensive Japanese sword doesn’t immediately make anyone a warrior. Join this interactive session to find out what it takes to become a truly data driven organization and leverage the potential of Big Data and Data Science
Networking Drinks Reception
Registration & Breakfast
How to Create a Successful Third Party API
Caroline will discuss lessons on API development, which led to Twitch's new efforts to develop a powerful and reliable third-party API.
Understanding Deep Neural Networks
Deep neural networks have been responsible for recent breakthroughs in speech recognition (Siri & Cortana), pattern recognition (self-driving cars), and machine learning (predicting NFL football scores). In this lively and informal session, Dr. James McCaffrey from Microsoft Research will explain exactly what deep neural networks are and how they work, without using Greek letters or annoying math jargon. You'll leave this session with a solid understanding of deep neural networks, what they can and cannot do, and have all the information you need to communicate with subject matter experts.
Data as Spaghetti: How to make Quality Content an Appetizing Business
We are in a world awash with content junk food: it's fast, fun, cheap - but not necessarily that good for you. As a dad, I would always hide vegetables in the spaghetti I made for my kids. Over time, this approach made spaghetti become a gateway to much healthier eating. Learn how data-driven products can make quality, award-winning content taste like sweet success.
Networking Coffee Break
Emerging Regulatory Challenges to Customer Insight & Personalization
Learn how recent developments in the law like the EU GDPR, the China Cybersecurity Law and the California Consumer Privacy Act may impact data science and analytics. We'll discuss practical steps you can take now to prevent programs like customer profiling, resolution and segmentation from being curtailed by legal requirements.
The famous astronaut, John Herrington once said: "there are no dreams too large, no innovation unimaginable, and no frontier behind our reach." There is one reason why every single one of us is in this room right now, and it's one-word "Innovation".
Innovation is an essential part of our daily life, and we are impacted by it every day. It is what allows the competitive advantage in all of our companies, and it's essential to meet consumers/customer needs. Today we're going to discuss the framework on how you can kickstart and lay the foundation that can lead and advance the culture of innovation within your company and help with the transformation.
Leveraging Data Innovation
During the last few years, being data driven is top priority
for organization success. Becoming data driven requires the right combination
of organizational data culture, governance process, and a scalable data
platform. With data technology landscape changing so fast organizations are
facing lot of challenges with business implementation and keeping up with
emerging technology innovation. During the session, we will present few
emerging trends in big data technology landscape, and possible approaches to
leverage the on-going innovation.
Panel Discussion: Creating Big Value from Big Data
While some deem “Big Data” as an overused buzzword, the need for large-scale data analytics has become crucial for organizations to tackle head on. With the multitude of information in today’s data landscape, the need to implement effective strategies in productive modeling, data visualization, text analytics, etc. has become more crucial than ever before. Join a panel of subject matter specialists to learn about today’s current trends & strategies in addressing today’s data overload.
• Driving holistic decision making with business analytics
• Promoting a proactive, innovative culture in leveraging big data to decision making processes
• Aligning your organization's strategy and long term goals to your data analytics roadmap
• Prioritizing data- effectively architecting our environment to give business units the data they need
o Historical data vs. streaming data (data coming in at real time)
• Translating data into actionable consumer insights and better decision making
o Ethics of data collection & protecting consumer privacy
• Utilizing today's latest technologies to translate data into organizational value
• Best practices in data science, predictive analytics, text analytics, etc.