Machine Learning: New Companion For Knowledge Workers

Jeroen Reizevoort

As machine learning gains ground thanks to smart algorithms and enormous amounts of data, working with an intelligent digital system may be less farfetched than you might think. Knowledge workers in particular stand to reap the benefits of this technology.

What is machine learning?

First, what exactly is machine learning? In essence, it refers to computer systems that can train themselves and get smarter. Data is their training material; the computer systems deduce knowledge from offered data, as opposed to data that is explicitly programmed. The more data is imported into a machine learning system, the smarter the system becomes. The use of a feedback loop is important in this regard: When wrong decisions are made, the system user indicates the correct answer, and the computer learns from its mistakes.

In daily life

Machine learning is not used only in specialized company software or labs. Smart computers are cropping up more and more in daily life. For example, consider the digital assistant on our smartphones (Siri, Cortana, Google Now), smart thermostats, and self-driving cars.

Until recently, computers excelled primarily at processing structured data—information stored in neat rows and tables. Thanks to machine learning, they are now capable of handling unstructured data such as speech, text, images, and videos. Computers recognize patterns in unstructured data and are able to draw conclusions from it.

As a result, machine learning is becoming more usable in the corporate world. Forecasts indicate that machine learning will grow by 50% per year until 2020. This will, of course, have a huge impact on the knowledge economy. Machine learning systems and knowledge workers will be able to work together more effectively.

Reaching better decisions together

Computers and factory workers have been working together effectively for years in production processes, and a comparable cooperation will emerge between computers and knowledge workers. People can focus on decision-making that adds value while the computer performs frequent and repetitive tasks. Together they will come to better decisions than people or machines can reach on their own.

Here’s an example from the medical industry: It is estimated that the body of medical publications doubles every 5 years, making it impossible for medical professionals to keep up. A trained system, however, can process published data and support physicians in their diagnoses. This is particularly valuable when it comes to rare diseases, and it also decreases the number of tests patients must undergo as physicians are able to exclude certain illnesses based on processed data.

The role of speech will become more important in the interaction between humans and computers. An example of this is the navigation of a management dashboard with the aid of Amazon Echo.

Computer vision

Another interesting development in machine learning is computer vision. In the 90s, a two-year-old child was able to provide a better description of an image than a computer. Now a computer can describe an image even more accurately than an adult.

The system is trained by importing a large amount of labeled images. The computer then uses this knowledge to recognize new images based on fuzzy logic (vague/woolly logic) algorithms.


Computer vision also has great potential in corporate life. A mechanic, for example, can take a picture of a defective part; the computer immediately recognizes it and orders a replacement. Knowledge workers can also benefit: A machine learning algorithm can analyze TV broadcasts of sports events and send marketing departments information such as when and how long their brand was aired (brand intelligence).

Why now?

Why is machine learning gaining momentum at this point in time? There are three important reasons:

  1. Big Data. Thanks to the digital economy, today we have more data—training material—than ever before. This data enables more powerful decision networks, with more decision variables. There is another side to the medal, however: Since more information is available about the very recent past, there is the risk that the most recent information will have a greater impact on decision-making. This is called the Recency Bias.
  1. Big computing. You get lot of horsepower for your buck these days. In particular, the chips in graphic cards (GPUs) are able to process large amounts of training data.
  1. Algorithms. The algorithms at the basis of machine learning have improved greatly in recent years. New algorithms have also made specialized forms of machine learning possible, such as deep learning and reinforcement learning.

A reflection of reality

For machine learning to work well, it is important that training material accurately reflects reality. This isn’t always the case. For example, the Dutch financial newspaper Financieel Dagblad recently reported on how a computer was used to judge a beauty pageant. Because the computer had been trained primarily using images of Caucasians, nearly all the winners were Caucasian and darker-skinned contestants were rated lower.

Machine learning will steadily become part of our daily lives. Keep watching this space.

For more on this topic, see Artificial Intelligence: The New Killer Feature.

About Jeroen Reizevoort

Jeroen has been working in IT for more than 25 years. In this period he has been active as a programmer, information analyst, project manager and, more recently, as pre-sales software architect. In this last role he has focused on integration, business process management, business rules management and mobile. Jeroen is currently working at SAP as pre-sales enterprise architect. He advises organizations on the definition of, and transition to, a modern, flexible architecture that is ready for the digital economy.