deep learning

Definition: Deep Learning

Deep learning is a machine learning technique that builds artificial neural networks to mimic the structure and function of the human brain. In practice, deep learning, also known as deep structured learning or hierarchical learning, uses a large number of hidden layers – usually more than 6 but often many more – of nonlinear processing to extract features from data and transform the data into different levels of abstraction (representations).

As an example, suppose the input data is an array of pixels. The first layer usually extracts pixels and recognizes the edges of image elements. The next layer can build simple elements from the edges, like leaves and branches. The next layer could then recognize a tree, etc. Data passing from one layer to another is considered a transformation, turning the output of one layer into the input for the next. Each layer corresponds to a different level of abstraction and the machine can learn by itself which data characteristics to place in which layer/level. Deep learning differs from traditional “shallow learning” because it learns much deeper levels of abstraction and hierarchical representation.

Why is Deep Learning important?

This learning technique is a revolutionary tool for processing large amounts of data, since the machine’s performance improves as it analyzes more data. As the amount of data increases, the machine becomes more capable of recognizing patterns, even hidden ones, among the data. As the machine also learns from the processed data, it is able to perform feature extraction and abstraction automatically from the raw data with little or no human intervention.

A lire également  Definition: Spotify Connect

Practical Uses of Deep Learning

Automatic speech recognition – All major commercial speech recognition systems (think your smartphone assistant) use some Deep Learning technique, with Recurrent Neural Networks currently being the most popular.

Computer vision – Images are used to train the machine to recognize features and now machines are showing “superhuman” accuracy in image recognition.

Natural language processing – Modern deep learning techniques have improved translation and language modeling. Google Translate uses deep learning techniques to translate based on the semantics of an entire sentence instead of just memorizing sentence-to-sentence translations.