Javascript must be enabled for the correct page display

“Separating Signal from Noise” explores the mechanics of machine learning at Signiant

Jun 15, 2020 By Ian Hamilton

Signiant’s CTO, Ian Hamilton, presents “Separating Signal From Noise” a discussion on artificial intelligence and machine learning, what can be learned from digital signal processing, and how machine learning adds business value today. The presentation is broken into the following five segments: Introduction, Neural Networks, Digital Signal Processing, Machine Learning and Digital Signal Processing, and Machine Learning at Signiant to help you better understand digital signal processing.

People vs Computers…How We Learn

While humans often learn best from analogy, computers excel at memorization and pattern recognition. The latter makes up much of the foundation of ML. In this introductory session, Ian describes how the video segments in this series leverage analogy and parallel perspectives as tools to build intuition into how machines learn.

Neural Networks

This second segment provides a high-level overview of neural networks–a common building block in most of today’s machine learning systems. This overview describes what neural networks are and the basic math behind them. People already familiar with these subjects might want to proceed to the following section.

Digital Signal Processing

This third segment provides an overview of digital signal processing. This overview describes digital signals and operations commonly performed on them. The simple math behind digital signal processing operations is also covered. People already familiar with Digital Signal Processing operations like correlation and transformation and digital filtering using convolution may want to skip this segment and go directly to the next segment.

Machine Learning and Digital Signal Processing

This fourth segment expands on the previous two segments to explain how machines learn.  This segment highlights similarities between neural networks and digital signal processing including why artifacts in these domains can be unintuitive. This segment establishes a basis for trusting neural networks.

Machine learning at Signiant

Signiant helps M&E companies leverage machine learning today both directly and indirectly, directly by leveraging machine learning in our patent-pending intelligent transport architecture. That architecture looks at a variety of inputs from both current and historical network conditions to determine the best possible path to move a particular data set over a network under the current conditions. Signiant also enables the use of machine learning indirectly. With our unique SDCX architecture, companies can leverage cloud ML and AI tools across multiple cloud providers without having to move the heavy assets into and out of the cloud.

Enjoy the video, and for those who would like to learn more about Signiant’s technology, please contact us to set-up a discussion.

Suggested Content


Is your company prepared to work remotely?

Any industry as vast as M&E is relatively unpredictable, and the surrounding unpredictable world only makes it more so. With remote access, media businesses remain prepared for any disruption that may come their way —whether it’s coronavirus or (hopefully) something less severe.


Is 8K More Supply or Demand?

As it stands, 8K’s efficacy is heavily intertwined with the nature of individual projects and requirements of the deliverables in question, and understanding the key strengths and current limitations of the format will be essential to its efficient and cost-effective implementation.


The Bandwidth Conundrum

On the surface, it may seem that moving large files from one location to another is relatively simple and that a faster pipe is all you need, but it turns out there’s much more to it. Learn why a bigger pipe doesn’t lead to faster file transfers and how to get the most out of your network.