81-83 Campbell Street, Surry Hills NSW 2010
Monday to Friday: 7AM - 7PM AEST
Artificial intelligence is intelligence demonstrated by machines, as opposed to natural intelligence displayed by animals including humans.
Artificial intelligence (AI) broadly refers to any human-like behaviour displayed by a machine or system. In AI’s most basic form, computers are programmed to “mimic” human behaviour using extensive data from past examples of similar behaviour. This can range from recognising differences between a dog and a car to performing complex activities including in the discovery of critical minerals that the world needs.
Deep learning is a field of artificial intelligence research that is concerned with the design, training, and application of neural networks to problems of such complexity as recognising facial features in images and recognising speech. Deep learning is characterised by its ability to learn complex representations that are capable of generalisation, abstraction, parallel structure and abstraction from small input data sets. The classic demonstration of deep learning systems came from image classification tasks such as image recognition or object detection. In these examples, the system learns to recognise features in an image or object in order to classify them as known objects or unknown objects.
In recent years, deep learning has been used for other applications including language processing, web crawling, personalization and fraud prevention. Deep neural networks (DNNs) have also been applied to tasks such as image captioning and sentiment analysis.
Deep-learning systems have been widely used in various applications since they were first introduced back in the 1990s. They have since become an important component of many advances made in machine learning and computer vision. A recent study showed that if you train a deep neural network with a simple text-classification task you can achieve a 95 percent accuracy rate.
Deep neural networks could also outperform state-of-the-art algorithms.
The recent explosion of deep neural networks has captured the attention of the scientific community and democratized artificial intelligence, but it is not clear whether these new approaches have the potential to improve a wide range of tasks.
Although we’ve been using deep neural networks for years, they have gained significant attention only recently. Deep belief networks are a recent development that make use of back-propagation learning to solve a variety of applications including image recognition, speech recognition and natural language processing. Convolutional neural networks use convolutional layers in several layers to learn features similar to those learned by humans. In both cases, back-propagation learning is used to update weights that control the activation function of each layer of the network.
In the ’90’s the field was dominated by simple convolutional networks that only provided landmark-like improvements in image recognition. In the past couple years things have moved on to more sophisticated deep neural networks, which are capable of learning seemingly complex features from restricted amounts of data.