5 December 2018
In this talk on databases powered by artificial intelligence, we will describe its design and implementation. Hint: we use unsupervised neural network approach - Word Embedding - from Natural Language Processing to build a model from relational data and then apply the model to enable standard SQL-based AI queries on relational data. Join us to learn more about our approach and the various use cases in the finance, insurance, and retail domains.
We will highlight unique capabilities of AI-powered databases by demonstrating different workloads using an Apache Spark 2.3.0 based prototype. We will also discuss key aspects of the Spark-based implementation, e.g., UDF implementations of various cognitive functions using Spark SQL, Python (via Jupyter notebook) and Scala based interfaces, Distributed Spark implementation, and integration of GPU-enabled nearest neighbor kernels.
Don’t miss this thought-provoking presentation and stay after the talk to connect with the presenter and other IBM researchers. Ask follow-up questions and/or how you might be able to implement similar strategies and approaches to your business. We look forward to hosting you!
Some experience required: the talk will include many references to SQL analytics, enterprise data/analytics, and big data. Thus, some experience or knowledge of these topics will be helpful in understanding this material. Whether you possess this experience or not, you are always welcome to come and learn!
Speaker: Rajesh Bordawekar, Principal Research Staff Member, Distinguished Scientist, and Master Inventor @ IBM.
Rajesh’s current interest rests in applying AI techniques for data management workloads and associated software-hardware co-design. He works at the intersection of high-performance computing, analytics, and data management domains. Recently, he has been investigating how GPUs could be used for accelerating key kernels in text analytics, data management, graph analytics, and deep learning. Rajesh has published over 40 technical publications and has been issued 20 patents.
- 6:30-7pm - Doors Open for Networking & Food
- 7-830pm - Presentation by Rajesh
- 830-9pm - Networking
- 9pm - Venue closes