In the accounting, audit, and compliance professions, Artificial Intelligence(AI) is already beginning to automate labor-intensive tasks, such as data entry or combing through manual documents. For example, accounting and audit professionals may deploy AI to extract information from invoices or purchase orders to enter into accounting and auditing systems, freeing up hours in the day to perform more robust analytics on the outputs of the system. Taking advantage of AI could enable accounting, audit, and compliance departments to analyze significant amounts of data, and deliver more analysis and insight as a part of their roles. In fact, AI has been making our lives easier for longer than we may realize. Your smartphone, your car, your bank, and the devices in your house may use AI on a daily basis to predict your preferences or anticipate your actions.
Space-time is a profound concept in physics. This concept was shown to be useful for dimensionality reduction. We give theoretical propositions to show that space-time is a more powerful representation than Euclidean space. We apply this concept to manifold learning for preserving local information. Empirical results on non-metric datasets show that more information can be preserved in space-time.
When businesses as influential and enormous as Microsoft and Google are rearranging their corporate structures around AI, it's clear the field is an important one. Artificial intelligence development is the next logical step in a world where large amounts of data are available. After companies started collecting ever-increasing amounts of information, they had to figure out what to do with that data and build the tools to help them do it. AI is often associated with automation, but it goes beyond the simplest, manual forms of that. For example, businesses could use AI to find patterns in the content they save to the cloud, and then focus time on putting that information to use instead of sifting through reams of data to find the insights that a machine can now highlight for them.
The mutual information is a core statistical quantity that has applications in all areas of machine learning, whether this is in training of density models over multiple data modalities, in maximising the efficiency of noisy transmission channels, or when learning behaviour policies for exploration by artificial agents. Most learning algorithms that involve optimisation of the mutual information rely on the Blahut-Arimoto algorithm --- an enumerative algorithm with exponential complexity that is not suitable for modern machine learning applications. This paper provides a new approach for scalable optimisation of the mutual information by merging techniques from variational inference and deep learning. We develop our approach by focusing on the problem of intrinsically-motivated learning, where the mutual information forms the definition of a well-known internal drive known as empowerment. Using a variational lower bound on the mutual information, combined with convolutional networks for handling visual input streams, we develop a stochastic optimisation algorithm that allows for scalable information maximisation and empowerment-based reasoning directly from pixels to actions.