Meetup #4

PLEASE NOTE: Limit of 100 attendees

Welcome to the next in the BD&ML Meetup series, and what we hope will be another interesting evening of presentations and networking

The agenda is listed below, followed by further details about the main presentations and their presenters.

Should you wish to contact me, email me at mark.whalley@microfocus.com.

Kindest regards
Mark

---

Agenda

18:30 Doors open and networking

18:55 Welcome
Mark Whalley

19:00 Running Machine Learning at Scale by Combining Python with Vertica
Badr Ouali

20:00 Teaching Computers to Play Games
Katie Scott

20:30 Beer, Pizza and Networking

21:30 Close

---

Running Machine Learning at Scale by Combining Python with Vertica
Databases leverage a range of programming languages. Data scientists prefer R and Python to build and test their Machine Learning models. And, they are not particularly interested in changing their tools. However, using R and Python alone does not address the Big Data opportunity to develop more robust Machine Learning models, based on the full corpus of data without down sampling. Join us to learn how you can take a best of both world’s approach by harnessing the power of the Vertica SQL analytical database and in-database Machine Learning capabilities with the vPython Library to develop, test, score, and perfect the end-to-end Machine Learning process at scale.

Badr Ouali
Badr, a Data Scientist, joined Vertica in November 2017. Prior to Vertica, Badr received both an undergraduate and Master’s degree in Computer Science/Mathematics from the National School of Computer Science and Applied Mathematics in Grenoble, France. Badr is passionate about sharing knowledge and insights about anything related to data analytics with colleagues.

---

Teaching Computers to Play Games
Until recently, storage and processing capabilities of computers were insufficient for training intelligent machines in any practical amount of time. In 2015, DeepMind’s AlphaGo was the first computer to ever beat a champion in Go – a game with a problem space many orders larger than chess. Then last summer, OpenAI created an agent that could beat the world’s top professionals 1v1 in the MOBA game Dota 2. Also during the summer of 2017, I created an agent that could beat humans at a brand new 1v1 strategy game where the aim was to be faster than your opponent by casting different abilities at the right time. I will cover how this agent was created using Reinforcement Learning techniques, some tips and tricks learned along the way, and my thoughts on the future of machine intelligence.

Katie Scott
After completing her bachelor’s degree in Computer Science in 2016, Katie joined Jagex as a graduate. She spent a year rotating around various tech teams until becoming a Data Engineer in the Analytics and Data Science department. Previous projects include a distributed multiplayer bumper cars game on the Oculus Rift, a compiler plugin to automatically detect and parallelize loops and a range of TensorFlow models. Most recently, Katie has worked on containerization technologies such as Docker, and is now facilitating the deployment and maintenance of all the models the Data Science team produce.