Reinforcement ML - Continuously Adaptive Convergence to Drive the Best Actions

Join us in July as we discuss reinforcement learning!

Agenda

6:00 PM -- Greetings, networking, food, and drink
6:30 PM -- Reinforcement Machine Learning - Continuously Adaptive Convergence to Drive the Best Actions -- John Hebeler
7:30 PM -- Closings and swag

Location
ClearEdge IT Solutions
10620 Guilford Rd Suite 200
Jessup, MD 20794

Location
Food and drink will be provided!

Talk
Reinforcement Learning (RL) is a very different learning beast. Rather than learning from supervised labeled data or unsupervised clustering, RL creates actions given an environmental context and then potentially receives a reward for the action’s results. Complex action sequences may be required to achieve a reward or punishment. As RL iteratively works to maximize its rewards, it learns more and more effective action steps including adaptation in very dynamic contexts. RL balances exploration via new actions and exploitation receiving the rewards from known actions to establish the optimum reward policy. RL offers tremendous value in dynamic, data-rich contexts such as games, robotics, complex human interactions, and more. RL has achieved super-human powers in several areas already. Through actual RL projects, you are guided to understand the principles and key programming techniques to dive into your own RL challenge. All code is available in a published GitHub repository. The practical projects cover both traditional and deep methods to train and test an RL model. Get started with RL!

Speaker
John Hebeler, PhD is a Lockheed Martin Fellow developing advanced machine learning systems to uncover anomalies and patterns-of-life from large, real-time data streams (Sensors, IOT, Cyber) and developing associated machine learning processes, tools, and strategies. Formally, he led a five-year program to analyze large, real-time data streams to form complex policy determinations in an event-driven architecture. John Hebeler holds six patents and is the co-author of two technical books and multiple journal articles on networking, data semantics, and machine learning. He has presented at major conferences around the world on NoSQL databases, machine learning, graph analytics, and large-scale, distributed architectures. He formally developed and taught graduate technology courses for the University of Maryland and Loyola University. John holds a BS in Electrical Engineering from Rensselaer Polytechnic Institute, an MBA from Loyola University, and a PhD in Information Systems from the University of Maryland, Baltimore County. In his free time, he’s an avid tennis player, beer brewer, and amateur audiophile, usually not simultaneously.