From journalism to marketing, generating creative content has increasingly been the target of stronger ML algorithms and growing data connectedness. Creative work is fast becoming a part of what machine learning can do well, and both creators and customers alike will benefit from a world where cutting-edge machine learning helps artists and authors reach their maximum potential.
Enter Rezrov Lab’s first offering in development, the Calliope™ engine. Rezrov Lab builds creative AI by applying cutting edge deep learning methods (e.g., hierarchical variational autoencoder networks) to generate natural language data a three-step process:
|The Calliope™ engine earns structure from unsupervised text inputs (e.g., stories, articles, novels)|
|The engine then builds novel content by recognizing existing patterns and tying new ideas together in similar ways|
|Finally, it “fills in the gaps” where the deep learning algorithm is uncertain by using statistical methods to turn its rough draft into fluid text|
Rezrov Lab is building Calliope™ as a B2C platform to generate its own new content as a marketable product. Our first commercial target is fiction — eBooks and articles. We look forward to completing our first implementation and perfecting an AI that helps us do what we’re best at — creativity.