Skip to content

Machine Learning
and Deep Learning

link|that opens up a new quality of solution possibilities with neural networks. But what are neural networks in AI and what is the difference between machine learning and deep learning?

Machine Learning (ML)

The basic definition of machine learning is:

Algorithms that analyze data, learn from that analysis, and apply what they learn to make informed decisions.

A simple example of machine learning is a streaming service for videos. An artificial intelligence is used to decide which additional videos to recommend to customers. This requires algorithms that use a learning process to capture the preferences of individual viewers and compare them to the behavior of other users. As a result, the AI predicts which films may also interest certain viewers and the result is played out in the form of film suggestions.

Due to the defined goals and the algorithm based on them, however, there is a natural limit for machine learning in terms of functions. Nevertheless, a multitude of application areas results:

  • Prediction of values based on the analyzed data.
  • Calculation of probabilities for certain events
  • Detection of deviations in existing patterns or according to certain specifications
  • Recognition of groups and clusters in a data set
  • Recognition of correlations in sequences
  • Optimization of business processes.

From automated malware scanning at IT security companies, weather forecasting to stock brokers looking for favorable trades. The algorithms are often designed to function like virtual personal assistants.

However, for machine learning to work and for artificial intelligence to make the decisions or predictions described, a human must program the logic of the algorithm. But things get really interesting when the computers themselves can make better and better predictions without being continuously trained and corrected by humans. In this case, we speak of deep learning and neural networks.

Neural Networks & Deep Learning

Deep Learning is based on Machine Learning and works in a similar way, which is why the two terms are often interchanged. However, the systems have different capabilities. Algorithms based on deep learning can independently improve the quality of their results. The system does not need to be explicitly programmed for its tasks; it can learn from training examples, for example.

The continuous analysis of data with a certain logical structure has parallels to human learning processes. The multi-layered structures of Deep Learning algorithms are therefore also referred to as “neural” networks. In relation to our streaming platform mentioned at the beginning, this means that the more detailed the AI analyzes the user’s behavior with the help of Deep Learning, the better or more accurate the recommendations will be – without a human having given the system clear logical guidelines.

What is a neural network?

The design of the neural network is based on the structure of the human brain. Just as we use our brains to identify patterns and classify different types of information, deep learning algorithms can be used to equip machines with similar skills. This makes deep learning much more powerful than machine learning. Whenever we receive a new piece of information, the brain tries to compare it to known objects. The same concept is used by deep learning algorithms.

Good examples of deep learning are computer programs that can play chess or Go. Both games require a clear mind and intuition. By playing against professional players, the algorithms determine the most efficient ways to succeed. The more games they finish, the more experience they gain. Learning success can even be achieved by pitting AI systems against each other. Through Deep Learning, the computers are eventually able to play at a previously unattainable level. These algorithms therefore manage to beat the best human players.

AI Solutions from link|that

AI solutions and neural networks play a central role in the link|that portfolio.

With link|that PRISM, link|that ECCO, link|that CarPark and link|that PowerMeter, we offer four very comprehensive standard solutions that deliver extremely reliable results. And even if you are looking for a very special solution for your company, our experts will be happy to assist you.

If you have any further questions or are interested in a specific solution from our range, we look forward to hearing from you.

Discover more in our Blog

Stories of our Customers

Want to find out more?