Feature Engineering for Hyper-Dimensional Decision Surfaces

IDEAS & Data Application Lab co-host this live webinar.

国际数据科学与工程协会 IDEAS

IDEAS is a global nonprofit organization that is dedicated to fostering the data engineering and data science ecosystems and broadening the adoption of their underlying technologies to accelerate the innovations data can bring to society. Our goal is to create a community to connect AI and Data Science enthusiasts. All of the conferences that IDEAS host will demonstrate cutting-edge technology and feature a variety of AI and Data Science experts covering topics including industry trends, real-world applications, open-source software, solutions-based case studies, and many others.

Guest Speakers:
Michael A. X. Izatt (LinkedIn Profile) is chief data scientist in the Insight & Analytics Group at Hitachi Consulting. Trained as a theoretical chemical physicist at MIT and Chicago, Max began his career with Bell Laboratories and Sandia National Laboratories in the Laser Projects Division of the Pulsed Power Research Group, where he reword essay on laser-initiated channel-ionization for space-based directed-energy delivery systems. He founded CentraLytics Corporation, a Microsoft Independent Software Vendor (ISV) at the ChicagoMercantile Exchange in 2002, which developed distributed-computing and collaboration pricing engines that coupled the desktop with the Microsoft enterprise business-intelligence stack for near-real-time high-frequency trading applications. CentraLytics’ intellectual property was successfully acquired within the Microsoft footprint in 2014. At Hitachi, Max works on Riemannian-motivated stochastic processes with thermodynamic applications in spin-glass and near-spin-glass convex systems, such as neural networks with applications in artificial intelligence applied to industrial process optimization.

Topics:
Feature engineering and physical analysis are integral to building predictive models for Industrial IoT applications. This presentation will review the background business case for hyper-dimensional decision surfaces, motivate the requisite mathematics, and present a formal definition of a Mercer Expansion for a kernel function and its certificate. A series of predictive classification models in R that demonstrate the predictive power of engineered features will then be reviewed.

This presentation is appropriate for quantitative analysts who work in engineering, finance, healthcare, pharmacology, manufacturing, retailing, distribution, and logistics.

About Data Application Lab:

About us

About IDEAS:

Home