Pocket Guide

Why Use Deep Learning for Pattern Recognition?

How Deep Learning Works

Deep learning uses neural networks to model the relationship between input data and output prediction. Think of the network as a series of layers, each performing specific operations. Early layers tend to look for smaller features within the larger pattern. Later layers look at feature combinations to determine the most probable data label.

neural network

Pattern Recognition

Deep learning is used to develop models that can find patterns in data. Non–deep learning methods, such as linear regression and k-means, tend to use more understandable logic and smaller amounts of simpler data.

linear regression

Complex Pattern Recognition

Deep learning is well suited for finding complex patterns. For example, you can perform object detection for tracking and localization or visual inspection. Complex patterns also exist in non-image data, such as audio and time-series signals.

  • Find and label specific objects in an image for object tracking and localization.

  • Visually inspect for defects and material damage.

  • Recognize spoken words or identify a song from a few-second snippet.

  • Find complex patterns in time-series signals for applications such as predictive maintenance.

Defining Features

Rule-based approaches require that you define all features. The difficulty is you must be familiar with the pattern to define the best features. You also need to handle all variations to avoid introducing bias. Complex patterns can make defining features time consuming or even impossible. With deep learning, pattern features are learned automatically.

Look for features in the acceleration data for high-five recognition.

Want to read a more detailed version?

Send to my email