The world is evolving towards an era where Artificial Intelligence (AI) will dominate the human race. An introduction to Deep Learning and how the technology is revolutionising the game-changing developments with massive amounts of computational power where machines can translate speech in real-time and recognise objects which the industries are already witnessing. AI is finally getting smart and will be a reality in the near future.
The term “Deep Learning” was first introduced to the machine learning community by Rina Dechter, a Professor of Computer Science at the University of California in 1986 and to Artificial Neural Networks – Artificial neural networks (ANN) or connectionist systems are computing systems vaguely inspired by the biological neural networks that constitute animal brains. Such systems “learn” to perform tasks by considering examples, generally without being programmed with any task-specific rules by Igor Aizenberg and colleagues in 2000, in the context of Boolean threshold neurons.
The first general, working learning algorithm for supervised, deep, feed-forward, multi-layer perceptrons was published by Alexey Ivakhnenko – was a Soviet and Ukrainian mathematician most famous for developing the Group Method of Data Handling (GMDH), a method of inductive-statistical learning, for which he is sometimes referred to as the “Father of Deep Learning” in 1965. A 1971 paper described a deep network with eight layers trained by the group method of data handling algorithm.
What is Deep Learning?
Deep Learning is a subset of Machine Learning in Artificial Intelligence (AI) that has networks capable of learning unsupervised from data that is unstructured or unlabelled. Also known as Deep Neural Learning (DNL) or Deep Neural Network (DNN). It utilises a hierarchical level of ANN to carry out the process of machine learning. People also refer deep learning to ‘Deep Artificial Neural Networks (DANN), and less frequently to ‘Deep Reinforcement Learning.’ The ANN is built like the human brain with neuron nodes connected like a web. The traditional programs analysed with data in a linear way. The hierarchical function of Deep Learning systems enables machines to process data with a non-linear approach.
According to Arthur Samuel, one o the pioneers in Machine Learning – a “field of study that gives computers the ability to learn without being explicitly programmed” – while adding that it tends to result in higher accuracy, require more hardware or training time, and perform exceptionally well on machine perception tasks that involves unstructured data such as blobs of pixels or text. Deep learning is a subset of machine learning, and machine learning is a subset of AI, which is an umbrella term for any computer program that does something smart. In other words, all machine learning is AI, but not all AI is machine learning.
The Important Features Of Deep Learning:
- Deep learning is a large neural network
- Deep learning is hierarchical feature learning
- It is scalable across various domains
- The deep stands for large artificial neural network
What does the term ‘Deep’ mean in Deep Learning?
Deep is a technical term that refers to the number of layers in a Neural Network. A shallow network has one so-called hidden layer, while in the Deep Neural Network, there will be multiple ‘hidden layers’ of neurons between these input and output layers, each feeding data into each other. Traditional neural networks contain only two or three hidden layers, while deep neural networks can have as many as 150 layers.
One of the popular types of DNN is known as the ‘Convolutional Neural Networks (CNN or ConvNet). A CNN convolves features learned with input data and uses 2D convolutional layers while making this type of architecture well-suited for processing 2D data, like images. CNN also eliminates the need for manual feature extraction. Hence you do not require to identify the features used for the classification of the images. The CNN works by extracting the features directly from the images. The relevant features are not pre-trained, but they are learnt while the network trains on a set of images. This automated feature extraction makes ‘Deep Learning’ models highly accurate for computer vision tasks like the classification of objects.
CNN learns to detect various image features using hundreds and thousands of hidden layers. Every hidden layer increases the complexity of the features of learned images. For, eg, the first opaque layer can learn how to detect the edges, and the last layer determines how to identify more complex shapes and sizes specifically to the shape of the object which you are trying to recognise.
Deep learning models are trained using large sets of labelled data and Neural Network architectures that learn features directly from the data without the need for manual feature extraction. Hence the term ‘Deep’ in ‘Deep Learning and ‘Deep Neural Networks.’ Multiple hidden layers allow Deep Neural Networks to learn features of the data in a so-called feature hierarchy. Networks with many layers pass input data features through more mathematical operations, than networks with few layers, and therefore are more computationally intensive to train. Computational intensity is one of the hallmarks of deep learning, and it is one of the critical reason why GPUs are in high demand for training deep-learning models.
Why does ‘Deep Learning’ Matter and How is it being used?
Deep learning can be applied to any data type. The data types you work with, and the data you gather, or any data you think of for the machine learning model to learn. Some of the data one can use are mentioned below.
- Sound (Voice Recognition)
- Text (Classifying Reviews)
- Images (Computer Vision)
- Time Series (Sensor Data, Web Activity)
- Video (Motion Detection)
For many tasks – for recognising and generating images, speech and language, and in combination with reinforcement learning to match human-level performance in games ranging from ancient Chinese Game – Go, to the modern such as Dota 2 and Quake III. In the accuracy of the word, Deep Learning achieves recognition accuracy at higher levels than ever before which helps consumer electronics meet user expectations. It is crucial for safety-critical applications like driver-less cars. Recent advancements in Deep Learning have significantly improved where it outperforms humans in specific tasks like classifying objects in images.
Deep Learning systems are the foundation of modern online-services that are used by Amazon to understand what we say – both our speech and the language we use to communicate with Alexa virtual assistant or by Google to translate the text when we visit a foreign-language website. Every Google search uses multiple Machine Learning systems to understand the language in our query through to personalising our results hence enthusiasts searching for ‘jazz’ isn’t inundated with results about guitars.
However, beyond these visible manifestations of Deep Learning – systems have started to find uses in every industry today. The applications include – Computer vision for driver-less cars, delivery robots and drones; language and speech recognition and synthesis for service robots and chatbots; facial recognition for surveillance in countries like China; helping radiologists to identify tumours in x-rays, assisting the researcher in spotting genetic sequences relating to diseases and identifying molecules which could lead to effective drugs in healthcare; allowing predictive maintenance on infrastructure by analysing IoT sensor data; underpinning the computer vision which makes the cashierless Amazon Go supermarket possible, offering reasonably accurate translation and transcription of speech for business meeting and the list is endless.
To summarise, Deep learning is the fastest-growing field in artificial intelligence, helping computers make sense of infinite amounts of data in the form of images, sound, and text. Using multiple levels of neural networks, computers now have the capacity to see, learn, and react to complex situations as well as or better than humans. Every industry will be impacted by deep learning, and many businesses are already delivering new products and services based on this new way of thinking about data and technology.
