Five Breakthroughs In Machine Learning Marketers Should Know About

Post written by

Jeremy Fain

Jeremy Fain is the CEO and Co-Founder of Cognitiv, the first neural network technology available for marketers.

There are many people who are using machine learning to advance their industries and push boundaries. For the purpose of this piece, we'll define a "breakthrough" not only as an advancement in technology or technique but also as an important development within a specific sector. With that in mind, here are five relatively recent developments that caught my eye:

1. Machine-learning software can now generate text that reads like it was written by a human.

While the opportunities for the future are limitless, OpenAI is the only player of its kind in the space currently. The company says its software has "the ability to generate conditional synthetic text samples of unprecedented quality" and can produce lengthy pieces of text based on a short sample that imitates the original writing style. So, if you give the algorithm a sentence or two that gives it context (i.e., location, subject, timing, etc.), it is able to unspool an entire narrative based on those sentences. Because there is a distinct possibility that this tool could be misused to create fake news stories or misleading or misattributed articles, OpenAI has said it will not make the most advanced versions of its model available to the public.

However, such tools could make it easier for companies to carry out content marketing strategies and produce articles that are consistent with their brand voice and guidelines. While it is obviously preferable to have a human do the job both from an ethical and quality control standpoint, there is no reason that such tools couldn’t be used to generate short blog posts (with oversight, of course) in order to beef up a company’s online presence.

2. Machine learning can analyze tweets to identify security threats and rate their severity.

The security company FireEye worked with researchers from the Ohio State University and the research firm Leidos to develop a new way of identifying potential software security vulnerabilities on Twitter. Their system combs through tweets to find mentions of vulnerabilities, and then uses a machine-learning algorithm and natural language processing (NLP) to determine how much of a threat each one is. They were not only able to predict which of the vulnerabilities reported on Twitter will be rated as "high" or "critical" with more than 80% accuracy, but also which ones will show up in the database.

Many brands use similar tools to monitor which topics are trending on social media, which then allows them to produce content that is relevant and topical. NLP can also be used to determine general brand sentiment, and to keep abreast of any potentially negative brand-consumer interactions.

3. Scientists have found a way to use machine learning on a quantum computer.

This is a huge breakthrough, despite the fact that quantum computers do not exist yet. Essentially, the current state of machine learning exists thanks to the recent explosion in computing power, which enables machine-learning algorithms to learn things faster and more comprehensively.

Quantum computers promise to be much more powerful -- and quicker -- than even today’s supercomputers, which means that if they were used to develop machine learning, they would be able to analyze data much more efficiently and on a larger scale, leading to more accurate algorithms (and better results for marketers). Thanks to this research, we now know that it will be possible to train machine-learning algorithms on quantum computers.

4. The health care industry is looking to machine learning to help with diagnostics and treatment.

This is not so much one specific breakthrough but a trend that has gained momentum in recent years. Machine learning’s ability to sift through gargantuan amounts of data and find previously unnoticed patterns has the potential to be a huge boon for the medical community (as well as other industries that rely heavily on data-driven decisions, including marketing).

Given the vast amounts of information that medical establishments already have on hand, and the life-saving benefits of an algorithm noticing something a doctor missed, machine learning has the potential to transform the diagnostic process. Breakthroughs have come in areas such as computer vision and natural language processing of electronic medical records. Companies like GoogleAmazon and IBM have also made their own contributions to this arena, each with varying degrees of success.

Beyond health care, this sort of diagnostic machine learning has myriad uses, from allowing hedge funds to identify promising stocks to invest in (similar to what JPMorgan is doing) to helping homeowners monitor water use and minimize damage from leaks to monitoring energy use to cut costs (as smart plugs do).

5. Machine learning can make 3D printing more accurate and cost-effective.

There are many industries where precision in manufacturing is crucial. In the automotive industry, for example, parts that are not manufactured according to specification can lead to injury and increased production costs.

Now, researchers have created a machine-learning algorithm that enables "improved geometric accuracy" by analyzing the product information and correcting design models. Not only does this improve the quality of the parts being manufactured, but it also reduces waste.

All in all, machine learning has been applied in myriad ways. But as with any technology, progress is not linear, so it will be fascinating to see what new breakthroughs will come to light over the next few years.

Forbes Agency Council is an invitation-only community for executives in successful public relations, media strategy, creative and advertising agencies. Do I qualify?
">

There are many people who are using machine learning to advance their industries and push boundaries. For the purpose of this piece, we'll define a "breakthrough" not only as an advancement in technology or technique but also as an important development within a specific sector. With that in mind, here are five relatively recent developments that caught my eye:

1. Machine-learning software can now generate text that reads like it was written by a human.

While the opportunities for the future are limitless, OpenAI is the only player of its kind in the space currently. The company says its software has "the ability to generate conditional synthetic text samples of unprecedented quality" and can produce lengthy pieces of text based on a short sample that imitates the original writing style. So, if you give the algorithm a sentence or two that gives it context (i.e., location, subject, timing, etc.), it is able to unspool an entire narrative based on those sentences. Because there is a distinct possibility that this tool could be misused to create fake news stories or misleading or misattributed articles, OpenAI has said it will not make the most advanced versions of its model available to the public.

However, such tools could make it easier for companies to carry out content marketing strategies and produce articles that are consistent with their brand voice and guidelines. While it is obviously preferable to have a human do the job both from an ethical and quality control standpoint, there is no reason that such tools couldn’t be used to generate short blog posts (with oversight, of course) in order to beef up a company’s online presence.

2. Machine learning can analyze tweets to identify security threats and rate their severity.

The security company FireEye worked with researchers from the Ohio State University and the research firm Leidos to develop a new way of identifying potential software security vulnerabilities on Twitter. Their system combs through tweets to find mentions of vulnerabilities, and then uses a machine-learning algorithm and natural language processing (NLP) to determine how much of a threat each one is. They were not only able to predict which of the vulnerabilities reported on Twitter will be rated as "high" or "critical" with more than 80% accuracy, but also which ones will show up in the database.

Many brands use similar tools to monitor which topics are trending on social media, which then allows them to produce content that is relevant and topical. NLP can also be used to determine general brand sentiment, and to keep abreast of any potentially negative brand-consumer interactions.

3. Scientists have found a way to use machine learning on a quantum computer.

This is a huge breakthrough, despite the fact that quantum computers do not exist yet. Essentially, the current state of machine learning exists thanks to the recent explosion in computing power, which enables machine-learning algorithms to learn things faster and more comprehensively.

Quantum computers promise to be much more powerful -- and quicker -- than even today’s supercomputers, which means that if they were used to develop machine learning, they would be able to analyze data much more efficiently and on a larger scale, leading to more accurate algorithms (and better results for marketers). Thanks to this research, we now know that it will be possible to train machine-learning algorithms on quantum computers.

4. The health care industry is looking to machine learning to help with diagnostics and treatment.

This is not so much one specific breakthrough but a trend that has gained momentum in recent years. Machine learning’s ability to sift through gargantuan amounts of data and find previously unnoticed patterns has the potential to be a huge boon for the medical community (as well as other industries that rely heavily on data-driven decisions, including marketing).

Given the vast amounts of information that medical establishments already have on hand, and the life-saving benefits of an algorithm noticing something a doctor missed, machine learning has the potential to transform the diagnostic process. Breakthroughs have come in areas such as computer vision and natural language processing of electronic medical records. Companies like GoogleAmazon and IBM have also made their own contributions to this arena, each with varying degrees of success.

Beyond health care, this sort of diagnostic machine learning has myriad uses, from allowing hedge funds to identify promising stocks to invest in (similar to what JPMorgan is doing) to helping homeowners monitor water use and minimize damage from leaks to monitoring energy use to cut costs (as smart plugs do).

5. Machine learning can make 3D printing more accurate and cost-effective.

There are many industries where precision in manufacturing is crucial. In the automotive industry, for example, parts that are not manufactured according to specification can lead to injury and increased production costs.

Now, researchers have created a machine-learning algorithm that enables "improved geometric accuracy" by analyzing the product information and correcting design models. Not only does this improve the quality of the parts being manufactured, but it also reduces waste.

All in all, machine learning has been applied in myriad ways. But as with any technology, progress is not linear, so it will be fascinating to see what new breakthroughs will come to light over the next few years.

Forbes Agency Council is an invitation-only community for executives in successful public relations, media strategy, creative and advertising agencies. Do I qualify?

Jeremy Fain is the CEO and Co-Founder of Cognitiv, the first neural network technology for marketers.