NXP 반도체가 eIQ AI·머신 러닝 개발 소프트웨어에 2가지 새로운 툴을 추가했다. 소형 MCU부터 MPU에 이르기까지 엣지에서의 AI 배포와 사용이 보다 용이할 것으로 기대된다.
▲Two new tools added to NXP Semiconductors eIQ AI/Machine Learning development software / (Image: NXP)
Supporting AI deployment and use across a wide range of edge processors
NXP Semiconductors has added two new tools to its eIQ AI and machine learning development software, which are expected to make it easier to deploy and use AI at the edge, from small MCUs to MPUs.
NXP Semiconductors today announced the addition of two new tools to its eIQ AI and machine learning development software, making it easier to deploy and use AI at the edge across edge processors.
eIQ Time Series Studio provides an automated machine learning workflow for MCU-class devices such as the MCX MCU portfolio or the i.MX RT crossover MCU portfolio, simplifying the development and deployment of time-series-based machine learning models.
GenAI Flow provides building blocks for Large Language Models (LLMs) that power generative AI solutions. Designed for use with NXP’s i.MX family of applications processor MPUs, the generative AI solution trains LLMs on specific contextual data, making it easier to deploy intelligence at the edge.
For example, an appliance equipped with an LLM that has learned the user manual can converse with users in natural language about how to access specific functions, perform specific tasks, and optimize use and maintenance.
Deploying AI at the edge offers a number of benefits, including reduced latency, improved user privacy, and reduced energy consumption. NXP’s eIQ Toolkit extensions support these deployments. Developers have access to a variety of model types, from generative AI to time-series-based models to vision-based models. Additionally, users can deploy different models across a wider range of edge processors.
“AI is at the core of a world that anticipates and automates based on user needs and wants, and it must be developed for practical edge deployments,” said Charles Dachs, senior vice president and general manager of NXP’s Industrial and IoT business unit. “NXP offers an unparalleled breadth of options across the full range of AI models and AI-enabled edge processors.”
“With this tool, we are able to deliver practical edge AI to developers across a broad range of markets,” he said. “It’s suitable for small AI models on MCUs like our MCX portfolio, crossover MCUs like the i.MX RT700, and larger generative AI models running on more powerful devices like our i.MX 95 applications processors.”
eIQ Time Series Studio simplifies and reduces the time required to develop and deploy time-series-based AI models. It supports a variety of input signals, including voltage, current, temperature, vibration, pressure, sound, and ToF (Time of Flight), as well as signal combinations for multi-modal sensor fusion. Developers can extract meaningful insights from raw data processed in time order through automatic machine learning capabilities.
It can also quickly build customized AI models based on performance, memory, flash storage size, and accuracy criteria. In addition, it provides a comprehensive development environment including data curation, visualization and analysis, model automatic creation, optimization, emulation, and deployment. Software developers can create optimized anomaly detection, classification, and regression libraries through an intuitive interface without requiring deep data science or AI expertise.
Generative AI applications can be accessed on edge devices using NXP’s GenAI flow. This software pipeline provides the ability to optimize generative models such as LLM. It also supports Retrieval Augmented Generation (RAG) to securely fine-tune models on domain-specific knowledge and personal data without exposing sensitive information to model or processor providers. By linking multiple modules in a single flow, LLM can be easily customized to fit the task and optimized for edge deployment using MPUs such as NXP’s i.MX 95 applications processor.