Speculative Computation: Application Scenarios

Abstract

Artificial Intelligence and Machine Learning have been widely applied in several areas with the twofold goal of improving people's wellbeing and accelerating computational processes. This may be seen in medical assistance (for example, automatic verification of MRI images), in personal assistants that adapt the content to the user based on his/her preferences or, to optimize query response times in relational databases and accelerate the information retrieval process. Most of Machine Learning algorithms used need a dataset to train on, so that the resulting models can be used, for example, to predict a value or enable user-specific results. Considering predictive methods, when new data arrives a new training of the model may be needed. Speculative Computation is a Machine Learning sub-field that seeks to enable computation to be one step ahead of the user by speculating the value that will be received to be computed. A change in the environment may affect the execution, but the adjustments are rapidly performed. This paper intends to provide an overview of the field of Speculative Computation, describing its main characteristics and advantages, and different scenarios of the medical field in which it is applied. It also provides a critical and comparative analysis with other Machine Learning