AI & Research

AI research, NLP, computer vision, deep learning

Natural language processing is an important part of artificial intelligence that allows computers to understand and communicate with human language via machine learning. It is the discipline that enables computers to recognise, comprehend and generate their own text and speech. It combines computational linguistics, the models of human language as well as statistical modelling, machine learning and deep learning.

Deep learning uses artificial neural networks to solve complex problems and is inspired by the human brain. Deep learning systems stack multiple processing layers to allow them to extract abstract features from large quantities of raw data, without the need for manual, rule-based programming. The input layer is the receipt of raw data, like words in a text or the pixels that make up the image. Each layer then processes this data, and transforms it into higher levels of abstraction, for example going from recognising edges, then understanding that edges form shapes and that a shape in this context is likely to be a face. The output layer is the final result, which could be the classification of what the image shows. Backpropagation is when the network corrects its own errors by comparing its output to the desired result.

Computer vision is the subset of AI that trains machines how to interpret and analyse the visual information contained in images and videos. It uses deep learning models and neural networks, and can automate tasks where the goal is to recognise things, detect faces and have spatial awareness. These are elements that are meant to replicate human sight. Among other fields, this technology is being used in retail environments to automate checkouts without the need to scan bar codes, and in healthcare settings to allow the early detection of tumours with a high degree of precision.