What’s¬†poppin’ everybody? Summer hit me hard and I constantly feel on fire! ūüĒ• So let’s¬†just use this¬†energy to sum up what was going on in¬†June.

Summary_Technologieengel

 How machines learn to learn

Who teaches the machines to learn? Responsible is a subsection of machine learning, with the cool term “Deep Learning.” The learning methods are based on the functioning of the human brain and should enable the AI to make their own decisions and forecasts. Therefore, this training method uses neural networks. Read the next paragraph to understand what neural networks exactly are. In short, however, such networks are built in layers that consist of a series of interconnected nodes (= neurons or units called).

In order to be able to use such a system effectively, in addition to the neural networks, further training methods are needed. They use and analyze large amounts of data. On the basis of this information in combination with a neural network, the system is able to link the learned with new contents again and again and thereby to learn repeatedly. Consequently, a machine with such a system is capable of making decisions, predictions and questioning them. Either decisions are confirmed or changed in a new learning process. Normally humans do not intervene during the actual learning process.

Neural Networks

Before we rush into this, here is a little reminder of what a neural network is all about. This term describes a collection of information processing units (neurons) arranged in layers that consist of a series of interconnected neurons. Mostly, the connection between the neurons consists of two layers, but sometimes also within a layer. Each neuron of one layer is always connected between the layers with all neurons of the next layer.

summary_Technologieengel

The information flow always starts at the input layer and flows over one or more interlayers, up to the output layer. Similar to the human brain, the output of one neuron is the input of the next.

Input layer

Ready, set, go! Here is the starting point of all information that has to travel through the neural network. The neurons at the beginning of this layer receive input signals and pass them on to the neurons of the first interlayer. In this case, the respective information is always transmitted by a neuron of the input layer to all neurons of the first intermediate layer.

Interlayers

As the name implies, there is at least one interlayer between the input and output layers. It is also called activity layer or hidden layer. The more interlayers there are, the deeper the neural network is. Hence the name “Deep Learning.” Such networks often have 150 or even more layers. Theoretically, the number of interlayers is unlimited, but each layer more means an increase in the required processing power.

Output layer

The last of all layers is located behind the intermediate layers. All neurons of the output layer are connected to all neurons of the last intermediate layer. Furthermore, here ends the flow of information in a neural network and the results of information processing are available.

Net neutrality

Regardless of the transmitter, receiver, location, content, service and application, all data streams through a network must be treated equally. Net neutrality therefore means that the web-providers have to treat all data equally and transfer it equally fast or slow. Data transmitted through the internet must not be discriminated. Let’s take the radio as an example: Complete net neutrality would mean that all transmitters transmit at the same speed and no transmitter is preferred or disadvantaged. If the transmission speed is throttled for station A, this must be done to the same extent for station B.

summary_Technologieengel

It does not matter whether it is a small regional station or a public broadcaster. Under European law, the provider must ensure that the data of all channels are transmitted on an equal basis and no transmitter is favored or disadvantaged. The infrastructure, the cable or the Wlan network thus provides the Internet-provider. All market participants must be able to use these without any differences in quality, regardless of their size or the type of content.

What about you?

Have you had a great beginning of summer? We hope you enjoy the hot weather as much as you enjoy reading about the stuff that was going on in June! Any questions? Leave a comment below or contact us on Facebook ‚ėÄÔłŹ.

 

Copyright © Bilder: Unsplash