Breaking Through the Machine-Learning Hype

Machine learning is one of 2017’s most used buzzwords, with tech companies across the board raving about its applications. Pick an industry, any industry, and you’ll surely see news around how machine learning is helping businesses achieve unseen levels of innovation. Unfortunately, more often than not these expectations are inflated. Gartner recently ranked machine learning as a buzzword at the top of their peak of inflated expectations. It’s understandable that it’s becoming harder to see where the practical applications of machine learning lie.

In essence, machine learning is a general term describing how systems can learn from feedback loops to improve their performance. The term encompasses a variety of different technologies, making it difficult for the average business to realise how machine learning can specifically help them. Here’s an overview of the technologies you should know about and how they can be applied to a variety of businesses.

Machine learning and business applications

Due to the wide array of AI and machine learning media coverage, there’s a common perception that machine learning is limited to groundbreaking new companies who have large amounts of money to spend. This doesn’t have to be the case: machine learning has been used in the past by smaller companies with less budget, or by larger companies but not in the kind of media-friendly ways you might expect.

An example of this is Airbnb’s use of machine learning. The company is always working to get more bookings, competing with the hotel industry on both price and experience. To do so, it has implemented machine learning processes which work in conjunction with data to make search results more personalised to the company’s customers.This has boosted online conversion rates by 1% – a huge amount considering online conversion rates are usually very low.

While companies like Airbnb, sitting on huge cash reserves may find it easy to invest and experiment with these technologies, there’s a popular perception that businesses with smaller budgets would find this difficult. Despite what you might think, machine learning has matured rapidly, with many core component technologies open-sourced or productised in ways that make them accessible to smaller businesses on tighter budgets.

How to be practical about it

So, what are these technologies, and how are they typically used? Google’s machine learning team has developed a technology called TensorFlow, which is a framework to implement machine learning at scale. TensorFlow, which was open-sourced in 2015, has allowed businesses of all shapes and sizes, from startups to established brands like Ocado, to use machine learning for their specific needs.

Naturally, Google isn’t the only company developing these technologies. Amazon are also competing by offering their own solution, Amazon Machine Learning. This is different to TensorFlow in various ways – with the former, you can build your own models and execute them against datasets wherever you like. AML, on the other hand, requires that you upload your dataset to Amazon and use their API to execute queries. AML is a hosted machine learning product, while TensorFlow gives you more freedom.

Microsoft is another player in this market, providing a service called Azure Machine Learning Studio. This product helps users to see their algorithms, executing them in their Azure Cloud. On top of these three big corporations, there’s a wide array of smaller startups providing machine learning services, among them BigML, MLJar and Algorithms.

The point here is that you don’t need to be an expert in the field, or have heaps of money to use machine learning. Yes, these tools will require some upfront investment in research from your developers. However they have also made machine learning something affordable and accessible to all sorts of businesses. It’s a trade-off I predict more and more businesses will find attractive in the future.

[easy-tweet tweet=”There’s a huge range of options available as the machine learning API ecosystem develops” hashtags=”MachineLearning, AI”]

Find what works for you

When it comes to picking a machine learning service for your business, you will need an open platform that is compatible and extensible with the best services. Remember – there’s a huge range of options available as the machine learning API ecosystem develops, but that doesn’t mean you should be overwhelmed. As long as you focus on your specific business needs, you should have no issue making a choice.

It’s also important to decide whether you want a ready-made machine learning ‘product’ to use or a set of APIs. Among the solutions which behave more like products are those available are Amazon Rekognition which provides image recognition, Amazon Lex, which focuses on chatbots, or Amazon Polly which works on text to speech. These should be easy to integrate and have many uses.

Another option is to build features into your products, which are powered by APIs powered by machine learning. One example of this is Google Cloud Video Intelligence API. This is a service that is currently in beta which can recognise objects in videos.  You can also build and train your machine learning models depending on your expertise. For this, you can use AML or TensorFlow.

Think carefully about your problem and what resources you have. Research and assess what will work best for you. Machine learning has a wide range of applications, which may make it sound daunting, but the market is getting larger all the time. The big players like Google and Amazon are investing a lot of time and thought into making machine learning more accessible for ‘regular’ businesses, so there’s been no better time to jump in and experiment.

David Mytton, Server Density

David Mytton is founder and CEO of Server Density, a scalable infrastructure monitoring software company. Server Density offers a SaaS product featuring the graphing, dashboards and low management overhead that modern businesses need. Server Density has more than 700 customers, including the NHS, Drupal, Firebox and Greenpeace, and has offices in London and New York.

AI Readiness - Harnessing the Power of Data and AI


Related articles

Is Poor Collaboration Hurting Your Team’s Productivity?

Regardless of how many days you believe hybrid workers...

AI Decodes Human Behaviour Like Never Before

In this exciting episode of "The AI Show," we...

The AI Show – Episode 7 – Martin Taylor

In this episode of the AI Show, host Emily...

Three ways to protect your business from cyberattacks

Keeping on top of cyberattacks in this current digital...

Data Sovereignty in AI: Five Key Considerations

Data is the most valuable economic asset for modern...

Subscribe to our Newsletter