By Dan Fischer

This is a two-part blog series and is an introduction to deep learning analytics focusing on fundamentals instead of a technical “deep dive”.

While you may not have heard of deep learning, you may have seen movies like Minority Report and iRobot. These movies show a futuristic view of society and how robotics and analytics could transform our society.

Another example is the recent development of companies like Waymo and other self-driving car companies, which you can learn more about over at https://torc.ai/, would be an example of the breakthrough involving deep learning and it isn’t going away, in fact, it is only becoming more coveted. In 2012, Google had two deep learning projects and as of late 2016, it had over 1000. Google has even created a specific team called the Google Brain Team developed for the advancement of AI and Deep Learning.

As advanced as deep learning is, there is still only a couple models of machines progressively learning, artificial intelligence and machine learning.

  • Artificial Intelligence is defined as a field or system that allows machines to perform tasks that machines typically cannot intelligently do without human interaction. The machines also have the capabilities to progressively become more independently intelligent over time.
    • Rule based system: Data and rules are inputted by “humans”. An example of this is, if this happens, then this, and this, that this is the result. This is the lowest example of AI.
  • Machine learning would be a subset of Artificial intelligence. At this stage inputted features helps the “machine” automatically learn without being programmed explicitly. Machine learning can be great when working within data science. Those who are looking to get a data science job will learn a lot about data sets and machine learning, from learning the top techniques, to how to validate and evaluate data science systems. If you are interested in learning more about data science jobs or machine learning, see more at – https://www.springboard.com/workshops/data-science-career-track/.
    • Statistical Learning Models: This sub section of Machine learning relies still on “Humans” to add data/features for the “machine” to learn from.
  • Deep Learning would be a subset of machine learning. At this stage the “machine” identifies patterns in data and extracts relevant features automatically using neural networks. See Botpress to learn more about what a deep neural network is and how businesses can make use of them in things like chatbots.
    • Representational Learning: A subsection of Deep Learning and doesn’t require “Humans” to learns and acquires relevant data. This is where features can be learned automatically.

Look for part two in the next couple weeks focused specifically on neural networks and their key applications and the correlation to deep learning.

About the Author:

Dan Fischer is a Strategic Solutions Executive at Paradigm Technology. Dan is focused on our financial and high-tech divisions and can be contacted at dan.fischer@pt-corp.com

About Paradigm Technology:

Paradigm Technology is a strategic consulting company serving the utilities, banking, airline, manufacturing and high-tech marketplaces. We utilize innovative business and technology solutions to help clients enable their digital transformation programs, and improve their Analytics, Cloud, Master Data Management, and Project Leadership solution delivery. Paradigm is ready to support you in your business journey. For more information about Paradigm Technology, email info@pt-corp.com or visit us at www.pt-corp.com.