Jaxon

Boston, MA

Jaxon uses AI to build custom training sets for document classifiers, self-adjusting to domain-specific use cases. Jaxon automation improves and augments sparsely-labeled training sets with: • Autonomous Labeling: Labels data at scale for supervised learning. • Ensembling: Intelligently aggregates heterogeneous labelers, generated heuristics, and existing machine learning models. • Data Grooming: Sorts, filters, and partitions data into learning-optimized minibatches. • Human/Machine Cooperation: Incorporates SME-provided knowledge about the data and domain. Compared to traditional human-powered data labeling approaches, Jaxon produces high-quality, learning-optimized training sets in minutes vs. months at a fraction of the cost.