The rise of AutoML

By Ben Avner, Co-Founder and CTO, Matchly.

The concept of machine learning first emerged when Alan Turing wrote an article about the ability of machines to achieve artificial intelligence. In 1957, Frank Rosenblatt designed the first neural network, called the perceptron algorithm. They are called neural networks because they are thought to be designed based on a simplistic way of how the brain works in order to process information. Although there were some real early applications for machine learning, such as the Madaline network, which could eliminate background echo from telephone lines, it would not return to the fore until vision applications emerged. by computer in 2012.

In 2012, AlexNet, a deep neural network designed by Alex Krizhevsky, scored 84% accuracy in Imagenet's image classification competition. The previous best result was 74%. It was then that the widespread adoption of machine learning to solve computer vision problems began. Deep machine learning quickly became the norm and outperformed humans in many tasks. Examples include Google's diabetic retinopathy and breast cancer projects.

ML works by feeding a neural network large amounts of data and making it learn patterns by adjusting the activation levels of neurons within the network. It can solve a wide variety of problems for many different data types.

There are many techniques for producing ML models. Some of these techniques include:

• Embeddings: A technique for taking sets of data and converting them from a high-dimensional space to a low-dimensional space. This allows us to take a very complex data set and make it easier to use.

• Linear Regression: A technique that allows rapid and efficient modeling of the relationship between a scalar response and one or more explanatory variables.

• Trees: A technique that uses a decision tree to represent how different input variables can be used to predict a target value.

• Neural Architecture Research: An Automation Technique...