What is Deep Learning?

Deep Learning has really hit its stride over the last 12 months. Hardware reaching the computing power they have, means thomachines are now able to process that vast amounts of data necessary to optimise Deep Learning, and do so within a time frame that is of value to organisations.

On this page we explain what Deep Learning is, how Deep Learning adds value to organisations, and how you can get started.

Deep learning is about taking pure data that has previously been interpreted by humans, corresponding actions, and training computers to perform that same task by drawing from examples.

How Does Deep Learning Add Value?

Deep Learning improves upon existing techniques by processing larger datasets to construct more accurate models. Areas where deep learning models works better than alternatives are:


Computer Vision  

  • Classification (identifying what’s in an image)
  • Localisation (identifying what’s in an image, and where it is)
  • Segmentation (identifying what’s in an image, and which pixels it’s in)
  • Generation (generating an image)

Natural Language Processing

  • Classification (labelling a given text)
  • Summarisation (generating a paragraph or sentence for a given document)
  • Language Modelling (generating text eg., a new article headline)
  • Questions Answering (Answer a user’s question from text corpus)
Figure 5


  • Speech Recognition (converting audio to spoken words)
  • Classification (identifying what’s in an audio sample)
  • Generation (generating music)

Time Series Data

  • Classification (identifying a pattern in time-series data)
  • Prediction (predicting the next value in time-series data)
  • Generation (generating or simulating time-series data)

These capabilities allow us to automate tasks which were previously very costly for humans to perform.

How to Get Started

There is no one answer for how an organisation can get started. We suggest you get stated on a small experiment centering around a problem that has the potential to have a sizeable impact on the business. Working through the following points will help guide towards.

What is your problem?

You can’t just buy a bunch of deep learning, drop it in the office and have it work it’s magic. Instead consider what problem you are wanting to solve. Define the problem. Think about how big an impact a solution could have on the organisation.

Off the shelf vs. custom solution​

There are a huge amount of solutions currently on offer. The big players are all offering off the shelf solutions that your team can begin to implement without necessarily requiring a team of Machine Learning Engineers on hand. This is naturally a more cost effective solution.

Custom solutions, come into play for a number of reasons, but the underlying link is uniqueness.

  • Your data set centers around proprietary assets
  • Your use case is novel
  • Your data set is highly sensitive, and requires a system to be built around it


There are a few factors to consider with data for you deep learning project, collection costs and quantity.

Collection costs

If you do not have a data set all ready to go, considering the cost of collecting data will be key to the go ahead of the project. It is also worth investigating ready made data sources available online both for purchase and for free


Deep Learning models perform better the more examples they are given to train with. For a computer vision project, a minimum would be hundreds to thousands of quality images. By quality images, we mean images that are relevant to the outputs of the project rather than high resolution.


As Paul from Nvidia explained in his presentation, Deep Learning is now a feasible option for organisations thanks to advances in computing power. However, to harness Deep Learning you need to factor in how to access this computing power

On premise GPU’s

Investing in your own GPU workstation is a great option, both for POC and long term projects. This option is especially useful when your data set is sensitive

Cloud GPU’s

Using cloud computing power is also an option, however it is generally more expensive. Utilise this option if you are experimenting on a lightweight product. This option is also attractive to those whose data is already hosted with the cloud provider.

Why now?

While the theory of deep learning has existed for many years, there has been a resurgence of activity as more powerful computer hardware (GPUs) and larger datasets needed to train models have become available.

As a result, deep learning methods have recently achieved state-of-the-art (in some cases human-level or better) results in a large and varied number of fields including image processing, natural language processing, time series modelling and predictive analytics, as well as game-playing and decision making.

Contact us

Now is your chance to take the opportunity to be an early adopter of this technology.