Sequential models are significant in the AI (Artificial Intelligence) realm, particularly used in fields that require the prediction of future sequences based on past ones. They process sequential data - where order matters, like a series of words in a sentence or a stream of audio or video. They play a critical role in Natural Language Processing (NLP), speech recognition, and time-series forecasting. With sequential models, neurons preserve information from inputs that appeared earlier in the sequence, adding a temporal dimension to the model. Further, unlike traditional models, sequential ones consider the 'context' or 'state' to affect future results, making them more effective for tasks underpinned by sequential patterns.
The specific use of Sequential Models in AI erupted during the 1980s but gained popularity in the late 1990s and early 2000s alongside the rise of deep learning and big data technologies.
Key contributors to the development of sequential models in AI include Juergen Schmidhuber and Sepp Hochreiter, who were particularly influential in the development of LSTM (Long Short-Term Memory), a type of Recurrent Neural Network which is a widely used implementation of sequential models. Other contributors also include Geoffrey Hinton, Yoshua Bengio, and Ilya Sutskever, who made strides in the practical application of these models in AI systems.