Is deep neural network supervised or unsupervised?

 

Deep learning is a subsection of machine learning and Artificial Intelligence (AI) that uses advancements in technology to allow computers to acquire data, knowledge, and experiences in the same way that people do. Deep learning relies heavily on statistical and forecasting models, both of which are components of data science. Deep learning has numerous advantages for data scientists who are in charge of gathering, analyzing, and interpreting large amounts of data. Deep learning makes this procedure simpler and more efficient. At its most basic, deep learning can be viewed as a method of automating predictive analyses. Deep learning algorithms are stacked on top of one another in a hierarchy of growing complexity and abstraction, whereas traditional machine learning algorithms are linear.

Deep learning uses supervised learning in situations such as image classification or object detection, as the network is used to predict a label or a number (the input and the output are both known). As the labels of the images are known, the network is used to reduce the error rate, so it is “supervised”.

In contrast, neural networks can be used to cluster pictures based on similarities. A neural network can be used to extract the features, followed by an unsupervised technique such as k-means clustering. A semi-supervised deep neural network is one type of neural network.

Furthermore, autoencoders are neural networks that can be used for image compression and reconstruction by using a latent space representation of compressed data; in other words, it outputs whatever is inputted. These self-supervised learning neural networks are autoencoders.

Finally, reinforcement learning with neural networks can be used, which was the technique used by DeepMind to win the game Go.

As a result, deep learning can be supervised, unstructured, semi-supervised, self-supervised, or reinforcement learning, depending on how the neural network is used.