FEDOT.INDUSTRIAL: FRAMEWORK FOR AUTOMATIC MACHINE LEARNING FOR INDUSTRIAL TASKS
SKU0045
р.
р.
DESCRIPTION
The framework is designed to automate the development of predictive models for technical systems throughout their lifecycle using evolutionary optimization methods. It supports tasks such as forecasting, classification, and anomaly detection for univariate and multivariate time series, as well as spatiotemporal fields of various natures. FEDOT.Industrial analyzes data reflecting the functioning of technical systems (time series, tabular data, images, text) and outputs an optimized and trained predictive model.
ADVANTAGES OF THE DEVELOPMENT
Ability to transfer neural network models to computing systems of different architectures (including adaptation and compression)
Composite structure allowing integration of data processing blocks specific to particular domains into the model
Automated solution for a wide range of AI tasks in industry
Acceleration of processes by 10–25 times
APPLICATION AREAS
Information technology
FEDOT+LLM: HYBRID AI SYSTEM COMBINING AUTOMATIC MACHINE LEARNING AND LARGE LANGUAGE MODELS
DESCRIPTION
FEDOT+LLM is a flexible interface between the user and AutoML, partially replacing AI developers. The user simply describes the task related to data processing in free form, and the system, through a dialog interface, will request additional information, clarify the task, translate it into understandable development terms, write the code, and interpret the results. This system is the first to use language models at multiple stages of machine learning: initial data collection from the user via a dialog interface, analysis of the obtained data, and result interpretation. AutoML is configured using generative AI based on LLM, and the originality of solutions is achieved through adaptive evolutionary software.