Some Notes about Deep Learning

Some notes about Deep Learning - Augmented Insights

Way back in September last year we wrote a short post about Machine Learning to provide some background about this approach to AI. To follow on,  here’s some notes about Deep Learning. We hope this helps you understand the key differences between it and Machine Learning.

In essence, Deep Learning is an attempt to impersonate a human brain’s workings. We should mention this is a rather basic approach in comparison to the capability of our little grey cells.

Deep Learning is a separate category under Machine Learning. But while it is closely related there are distinct differences between the two approaches. For a start, with Machine Learning, data is processed through a single algorithm layer to arrive at pre-determined results. Deep Learning improves on this by using multiple layers to refine its output, improving output accuracy.

Machine Learning also requires data that is structured and contains labels. It can use unstructured data, but it will undergo some work to structure it before use. By contrast, Deep Learning processes unstructured data and, through training, build its knowledge of structures within the data provided. This process removes the need to manually state features within the data, which Machine Learning would need to work.

A Practical Example of Deep Learning

Let’s have a practical example; You want to identify trees in images. Deep Learning would process the training data and identify the features that make up a tree. It will take the information within the pixels that make up part of the tree, such as leaves and branches and so on, and classify them so that it can then take test data and then any future data to correctly tell a tree from a bus or a guitar or an elephant.  One thing to bear in mind is Deep Learning does need a large volume of data to get to that stage. But the fact it can do this without manually needing human intervention in the form of labelling images, and the multiple processing layers it uses, means it will ultimately be a more accurate method than Machine Learning.

So there you go – some notes about Deep Learning. Hope you are enjoying our snippets around the world of Artificial Intelligence. If you have any burning questions feel free to drop us a message and we’ll see if we can drum up some stuff to help you.

Leave a Reply

Your email address will not be published. Required fields are marked *