The rise of machine learning and how to make it a success

By Johan van Rooij

One phrase we often hear bandied around almost interchangeably with artificial intelligence and data science is machine learning. Machine learning is a set of techniques where we let a computer learn from examples fed to the solution in the form of data.

These machine learning techniques are behind the technological advances that we currently term AI, and a big driver behind AI increasingly becoming a ‘system technology’ i.e. a technology that becomes fundamental to life, not unlike electricity, the combustion engine and the computer in the past.

The Evolution of Machine Learning

Machine learning has been around for much longer than you might think. The term was coined by Arthur Samuels (IBM) in the 1950s while he was working on a program that could play checkers.  Arthur included learning techniques in his artificial chequers player that meant it could learn from games it played against itself, against human opponents and from book games. Arthur himself had taken inspiration from a model of human brain cell interaction, created in 1949 by Donald Hebb in his book ‘The Organisation of Behavior.’

Fast forward 70 years and we’ve seen machine learning do a lot more than just helping computers win games. Although Watson winning Jeopardy in 2011 and DeepMind’s AlphaGo beating the world’s Go champion in 2016 were memorable achievements.

Where we are now and where ML is going

Machine learning is in use in myriad ways and is already enabling technologies as diverse as driverless cars, fraud detection, product recommendations, identifying damage in images, enabling dynamic pricing and the large language models that power generative AIs like ChatGPT and Bard use.

While we are still some way off an artificial general intelligence that could learn to apply itself to as many diverse problems as the human brain, improvements in machine learning techniques and processing power means it will be applied to larger and more complex problems and will continue to change the world we live in.  

Success factors in machine learning projects

The success of machine learning in business settings is dependent on a number of important practices:

  • Strong domain knowledge and AI expertise: although machine learning is about learning from data, making the right decisions in designing a full-scale machine learning solution can only be done with intricate knowledge of the domain and business setting it is applied to.
  • Data quality and availability: learning can only be successful if you have sufficient data related to the problem that needs to be solved. You need to make sure the right processes and technologies are in place to ensure availability, accuracy, completeness, consistency and neutrality.
  • Business integration & change management: machine learning solutions need to work in the business environment they are designed for, so change management and tight integration is vital.
  • Depth and breadth in technical skills: being successful with machine learning requires different skills ranging from data analysis, data engineering and deep mathematical skills, to project management, change management and business integration skills.
  • Compliance and ethical knowledge and experience: privacy, GDPR, the effect of biases in data on outcomes and the ethical implications are all important considerations for a machine learning project.

Machine learning has great potential to fundamentally change business for the better. But it needs to be approached carefully, taking every angle into account. Success is dependent on a holistic approach.

Machine Learning at Valcon

At Valcon, we are leaders in machine learning implementations. We have deployed machine learning in numerous applications including forecasting, credit scoring, competitive pricing, and predictive maintenance. We also excel in AI in natural language understanding – combining intent recognisers with large language models to power modern AI solutions for customer service applications (chatbots, email automation, call analytics, call summarisation) and document mining and classification applications.

Want to know more? If you would like to discover how Valcon’s specialist AI team can support your business, please contact [email protected] and we’ll be in touch right away.