Glossary
This glossary offers a selection of key definitions for terms used throughout the Neuralk documentation.
Tabular Data
Tabular data (or structured data) refers to information organized in rows and columns, such as spreadsheets or databases. Each row typically represents a single record (e.g., a client, product, transaction, or event), and each column represents a feature or attribute (e.g., age, date, category). One important aspect of tabular data is the relationships and correlations that can exist across rows and columns, which are critical for extracting insights and building predictive models.
Tabular Foundation Models (TFMs)
Tabular Foundation Models (TFMs) are a new generation of frontier AI models capable of making predictions on tabular data for a variety of ML tasks (classification, regression, time-series) instantly and without any additional model training. At Neuralk AI we are developing powerful TFMs for industry-grade applications.
NICL (Neuralk In-Context Learning)
NICL (Neuralk In-Context Learning) is the core proprietary foundation model developed by Neuralk AI. It leverages an in-context learning architecture to provide state-of-the-art predictions on tabular data instantly and without additional model training.
In-Context Learning
In-Context Learning (ICL) allows models to adapt to new tasks by observing examples or prompts, without requiring explicit retraining. This approach is especially valuable in industrial settings, where labeled data is often scarce or changes rapidly.
Context
It refers to the surrounding labeled data or examples that a model uses to make predictions. For NICL, the context of labeled examples acts as a description of the task at hand. The model uses this contextual information to infer relationships and predict outcomes for new samples. In doing so, it effectively adapts to new datasets without further retraining.
Expert AI Modules
The Neuralk AI expert modules are specialized workflows provided by the Neuralk API, designed to handle complex business use cases from start to finish. They enable users extend the foundation model's capabilities and solve business use cases with complex logic out of the box.