What Deep Learning Can Do For Business
Every so often a new technology buzzword appears, to be picked up and repeated ad infinitum in presentations, pitches and articles just like this. From big data to blockchain, they are convenient selling tools, a necessary shorthand; but all too often our knowledge of what they actually refer to is only skin-deep. And there are few terms more mysterious to the uninitiated as deep learning.
deep learning market is predicted to grow rapidly The problem is that to use these technologies effectively, or even develop a strategy around them, we need to fully understand their nature and their capabilities before we start. The deep learning market is predicted to grow rapidly in the next few years to reach $1.7 billion by 2022, fuelled by growing usage across a wide range of industries. But why is deep learning expected to make such an impact? What exactly is deep learning, and how can it be applied in enterprise to produce tangible benefits? Read on to find out.
identifying tweets related to environmental disasters First of all, let’s be clear exactly what we’re talking about. Machine learning is a field of artificial intelligence that enables computers to learn without being explicitly programmed, simply from the data we supply it with. Clearly, an algorithm which can improve its performance without human intervention is extremely powerful, and those machine learning algorithms are currently used for a whole range of applications, from sorting your emails to identifying tweets related to environmental disasters
One type of machine learning algorithm uses neural networks, artificial neurons that are connected together and organized into layers. A neural network is designed to classify information in a similar way to the human brain, making decisions and predictions about the data it receives together with a level of probability. Based on whether those decisions and predictions turned out to be right or not, algorithms modify connections in the network, improving the classification performance.
Deep learning is a type of machine learning which uses large neural networks with many hierarchical layers, hence the ‘deep’ in the name – in fact deep learning is often referred to in the scientific community as ‘deep neural networks‘. Neither the concept nor many of the algorithms are new, but the implementation of deep learning has only recently become practical. Not only does it require very large amounts of data to perform well, but neural networks are also very computationally expensive, so it was only the advent of big data along with improvements in processing power that made it possible.
Different types of machine learning algorithm have their own strengths and weaknesses, but in general they excel at pattern recognition, leading to many useful applications such as computer vision and natural language processing. Until recently however, machine learning algorithms required training data to be labeled – i.e. pictures of dogs had to be labeled ‘dog’, so that the algorithm knew whether or not it had classified the picture correctly. This is known as ‘supervised learning’, and while it is fast and doesn’t require too much processing power, manually labeling the data beforehand is time-consuming and expensive.
deep convolutional neural networks‘ perform very well in visual recognition tasks But because deep neural networks employ multiple layers of learning, they are able to classify objects or words without being told if their previous classifications were correct. They identify more and more detailed features at each layer, and each layer learns from the one before it. This automatic encoding of features, without labeled data, is known as ‘unsupervised learning’, and it is key – the ability to use unstructured training data is of great benefit in real-world applications because there is now a huge amount of available training data out there.
