An Overview to Cloud AI
Google Cloud AI aims to democratize AI by helping businesses generate actionable intelligence from data to create customer value. No matter where you are in your ML journey, join us for an overview of Cloud AI to see how it can provide you with the capability to move from data to insight to action, using pre-trained machine learning models, tools to generate and host your own models, and out-of-the-box ML solutions.
Reinforcement Learning for Recommendation system
Recommender systems are crucial for online services to match users with personalized and contextualized content from a large collection of items. Most existing systems are based on supervised learning and are inherently myopic—estimating a user’s immediate response to a recommendation without considering the long-term impact on subsequent user behavior. Recently, reinforcement learning has shown promise in several fields like games and robotics. However, applying existing methods to large-scale production recommender systems with millions to billions of users and items involves many challenges.
In this talk, we'll discuss how to use reinforcement learning for large-scale sequential recommender systems—where the systems learn from trajectories of sequential interactions with users, and take a sequence of actions to optimize for long-term values, not just immediate rewards—all while satisfying real-world training and serving constraints. We'll present recent research results, and discuss future opportunities for improvement.
What's new in TensorFlow?
TensorFlow is an end-to-end open source platform for machine learning that makes it easy to design and deploy machine learning models from research to production. This year we announced an alpha release of TensorFlow 2 focusing on simplicity and ease of use, with updates like eager execution, intuitive higher-level APIs, and flexible model building on any platform. In this talk you will learn about TensorFlow, some of its new features, and how it is being used for research and real-world applications.
LSTM-Based Online Handwriting Recognition
Google launched Google Handwriting Input in 2015, which enabled users to handwrite text on their Android mobile devices. Since then, progress in machine learning has enabled new model architectures and training methodologies, allowing us to build a new machine learning model and reduces error rates substantially compared to the old version. We will give an overview on the latest released "Fast Multi-language LSTM-based Online Handwriting Recognition" and explain how it works.