Once the sole purview of academics and a few of the largest high-tech companies, deep learning now represents an approach that’s poised for rapid and widespread growth across a range of companies and industries.
Artificial intelligence (AI) was a concept that was introduced in the 1950s. Initially, AI was inherently a rules-based approach. Developers would create rules that would be used to process information. The theory was that if you wrote enough rules, you’d ultimately have an AI system that could work on its own to solve complex problems. However, over the years, the promised benefits of these rules-based theories weren’t realized in practice. It was simply too difficult to write enough rules, with the details required, to generate the intended results.
This paper offers an introduction to deep learning, providing an overview of how it has evolved and its key building blocks including using web data to power deep learning.