Proprietary Hardware-aware AI Model Optimization Platform

[INQ. NO. 2208E27] Nota AI (Nota Inc.) is a tech startup specializing in AI model optimization. Nota AIs aim is to bring the benefits of AI to everyone and everywhere, spreading its use across various industries and applications.

Nota AI’s main product is NetsPresso — a proprietary hardware-aware AI model optimization platform. NetsPresso enables application of AI solutions for a diverse range of industries including intelligent transportation systems and driver monitoring systems.
Nota AI maintains partnerships with global market leaders including ARM, NVIDIA, Samsung SDS, LG CNS, Microsoft, and AWS.

NetsPresso : Proprietary hardware-aware AI model optimization platform
NetsPresso is Nota AI’s proprietary hardware-aware AI optimization platform that automates the development process of lightweight AI models. The platform automates repetitive development steps, significantly reducing the time and resources required to develop an AI model and optimizes it for the target device to enable more efficient and effective deployment in production. This helps resolve AI-related constraints in business use cases by removing the requirement for high-spec servers and network connectivity while preventing high latency and personal data breaches. The resulting benefits are cost-efficiency and enhanced performance.

NetsPresso is modularized into three modules
One of the best features of NetsPresso is its modularity. The platform is modularized into three modules: model searcher, model compressor, and model launcher. Model searcher automatically searches optimized models for your target device. Model compressor offers optimal tradeoff between performance and efficiency. Model launcher accelerates models and simplifies deployment.

NetsPresso AI development steps users guide
The modularization of NetsPresso enables five different use scenarios. Each module can be utilized individually or in succession according to one’s needs. A user who has only a dataset can search a model via model searcher, compresses it using a model compressor and deploys the model using a model launcher.

A user who already has a model, on the other hand, may skip Model Searcher and begin the process with model compressor to compress the model that they already have. A user who has an optimized AI model can go straight to the model launcher. The commercial launch of NetsPresso 1.0 is scheduled for August 2022.

 
 
korean-electronics.com | Blog Magazine of korean electronics, brands and Goods