Machine Learning, Your Business
Curious what Machine Learning is, and what it can do for your business? As I prepare to ready a new Machine Learning solutions agency - morethan.software, I've been investigating some of the interesting projects being worked on in this field. I know, before I learned about the field, I had no idea the potential value this type of 'system' can have for business, health, and security alike. I thought I'd share a little insight I’ve picked up along the way, and try to put it into the perspective of a business owner.
Let's start with a subset of Machine Learning called Natural Language Processing (NLP). This is the very subject that got me into the field, due to its applications in Marketing.
What is NLP?
Well, in practical terms, its making the computer understand human languages. Typically, in the form of written text, an NLP system can do anything from summarize a large article into just the bare essential facts, to generate an entire article. It can be used to understand questions, and automatically generate responses - for use in support... or used to classify news or text in terms of topic, relevance to a subject, and general sentiment - while extracting statistics.
What makes it really machine "learning?"
There are two kinds of machine learning - unsupervised and supervised. As you may have guessed, this refers to whether an admin or user is monitoring and providing feedback to the system or if it carries on with its own assumptions. Machine "learning" system employ complex systems called "neural networks" to gauge, from previous input/output, what the output for an “input” should be. In other words, it gathers experience, and adapts future responses - just like we do.
What are some practical applications of Machine Learning?
There are some great articles on this subject for example, this one from Tech Crunch which lists some applications, like filtering garbage comments/spam and pushing relevant user content to the top, or pushing the 'right' products to a customer looking at an online shop, and even understanding an issue a user is trying to report - and finding a solution for them.
Popular applications of machine learning jump out of the Natural Language Processing world, such as illustrated in this Observer article, and into what is known a "Deep Learning" - where computers start to "recognize" entities within noise... such as faces in a picture, or your mouth to make that silly tongue stick out on snapchat - this creates the opportunity for a whole new level of personalization and interaction. With systems, able to 'recognize' you, your gestures, your mood, your traits - and analyze patterns of behavior in relation to stimuli (ads, products, web page layouts, sport maneuvers) - everything you touch and interact with can adapt itself to what will "feel" right, and be most relevant to you... while giving the system admins exactly what they need to know: who cares, what do they care about, and what will they do with it. This translates into way more sales, and much less costs.
Another great article from THENEWSTACK discusses utilizing machine learning to make judgement calls from real-time and historical data on the fly - often being able to make better 'business' decisions than lesser experienced human analysts, due to the ability to assimilate vast simultaneous, and huge historical data streams.
So, what can it do for MY business?
All Machine Learning applications and tailored to a specific set of goals. If you sell online, Machine Learning can be used to discover which products sell the most - which products should be bundled together (or at least be made visible to the same shopper) and automatically respond to those "FAQ Style" questions that users are so reluctant to look up themselves. Even if you don't sell online, analyzing your financial tables and scouring social networks can provide insight into where your best customers come from, and how to reach them more easily... or add an ML based lead nurturing solution and watch it filter out the tire kickers and bring the serious opportunities to your direct attention.
There are so many more applications and examples of using ML that I have no choice but to continue this post as a series - so keep an eye on my blog (ericmacdougall.com) and LinkedIn, and we'll jump into some real use cases, and open source examples that you can try yourself!