Gartner Hype Cycle for Emerging Technologies 2016: Deep Learning Still Missing

gartner_emerging-tech-hc-2016

For the 22nd year, Gartner has released its much-discussed hype cycle report on emerging technologies, “providing a cross-industry perspective on the technologies and trends that business strategists, chief innovation officers, R&D leaders, entrepreneurs, global market developers and emerging-technology teams should consider in developing emerging-technology portfolios.”

Reacting to last year’s hype cycle report (see below), I made the following comment:

Machine learning is making its first appearance on the chart this year, but already past the peak of inflated expectations. A glaring omission here is “deep learning,” the new label for and the new generation of machine learning, and one of the most hyped emerging technologies of the past couple of years.

gartner-hype-2015

Source: Gartner, August 2015

This year, Gartner has moved machine learning back a few notches, putting it at the peak of inflated expectations, still with 2 to 5 years until mainstream adoption. Is machine learning an emerging technology and is there a better term to describe what most of the hype is about nowadays in tech circles?

Machine learning is best defined as the transition from feeding the computer with programs containing specific instructions in the forms of step-by-step rules or algorithms to feeding the computer with algorithms that can learn from data and can make inferences “on their own.” The computer is “trained” by data which is labeled or classified based on previous outcomes, and its software algorithms “learns” how to predict the classification of new data that is not labeled or classified. For example, after a period of training in which the computer is presented with spam and non-spam email messages, a good machine learning program will successfully identify, (i.e., predict,) which email message is spam and which is not without human intervention. In addition to spam filtering, machine learning has been applied successfully to problems such as hand-writing recognition, machine translation, fraud detection, and product recommendations.

Indeed, machine learning has been around for quite a while. In 1959, per Wikipedia, Arthur Samuel defined machine learning as a “Field of study that gives computers the ability to learn without being explicitly programmed.” Yes, 1959—not exactly what one would call “an emerging technology.”

“Artificial Neural Networks” and “Deep Learning” (and variations thereof) are the most hyped buzzwords today, more than any other tech buzzword, I would argue. They  have also been around for a long time, but advances made by The Canadian Mafia (and others) over the last decade in training computers with big data using specialized processors have generated “the latest craze.” The tipping point(s), to borrow another buzzword, came in 2012 when two much-publicized breakthroughs occurred: The Google “Brain Team” has trained a cluster of 16,000 computers to train itself to recognize an image of a cat after processing 10 million digital images taken from YouTube videos. In the same year, a deep neural network achieved 16% error rate at the annual Imagenet Large Scale Visual Recognition Challenge (ILSVCR), a competition where research teams submit programs that classify and detect objects and scenes, a significant improvement over previous results. The rapid progress this method has exhibited over the last four years has prompted countless PhD candidates to switch the focus of their research to the newly promising field, the application of deep neural nets to other areas requiring the processing and analysis of unstructured data, vastly increased funding from VCs, and a fierce competition for artificial intelligence startups.

Deep learning is not the only approach to machine learning being pursued today—Pedro Domingos suggests five fundamental approaches, but the variations and combinations are numerous. It gets most of the attention today, particularly after the triumph earlier this year of Deep Mind Technologies neural net over one of the best Go players in the world, but other approaches (or a combination of approaches) may prove even more promising or more successful in solving other types of challenges. So maybe using “deep learning” in a hype cycle chart, regardless of the hype, is indeed not a good idea. Instead, maybe Gartner should have used “advanced machine learning” to describe the emerging technology or technologies that are at the core of the hype, excitement, and interest in machine learning (or “artificial intelligence” which is also an old term and very ambiguous one).

Here’s a definition of Advance Machine Learning:

In advanced machine learning, deep neural nets (DNNs) move beyond classic computing and information management to create systems that can autonomously learn to perceive the world, on their own. The explosion of data sources and complexity of information makes manual classification and analysis infeasible and uneconomic. DNNs automate these tasks and make it possible to address key challenges related to the information of everything trend.

DNNs (an advanced form of machine learning particularly applicable to large, complex datasets) is what makes smart machines appear “intelligent.” DNNs enable hardware- or software-based machines to learn for themselves all the features in their environment, from the finest details to broad sweeping abstract classes of content. This area is evolving quickly, and organizations must assess how they can apply these technologies to gain competitive advantage.

Excellent definition (and advice), distinguishing what is “emerging” from “artificial intelligence” (coined in 1955 to describe human-like intelligence displayed by a computer program) and “machine intelligence. Oh, the source of this excellent definition of “Advanced Machine Learning” emerging technologies? Check out Gartner Identifies the Top 10 Strategic Technology Trends for 2016, published in October 2015.

In the press release accompanying the new hype cycle chart, Gartner states categorically:

Smart machine technologies will be the most disruptive class of technologies over the next 10 years due to radical computational power, near-endless amounts of data, and unprecedented advances in deep neural networks [italics mine] that will allow organizations with smart machine technologies to harness data in order to adapt to new situations and solve problems that no one has encountered previously.

Originally published on Forbes.com