Nota Inc., a South Korean automatic AI model compression platform provider, participated in Arm’s AI Virtual Tech Talk series. Given the increased demand for AI technologies, Nota develops on-device AI solutions for varied everyday applications.
Providing Optimized AI Solutions
Arm AI, an AI hardware and software solutions developer, and its partners would host the Arm AI Virtual Tech Talk. Arm and other participating AI companies would discuss the newest AI practices, technologies, trends, and more during the event.
Besides promoting AI-related developments, Arm’s tech talks encourage the AI ecosystem to provide specialized solutions, meeting users’ AI deployment needs. Nota is also the first South Korean company to become an AI Virtual Tech Talk contributor.
Last year, Nota joined the Arm AI Partner Program, which offers partners access to high-class resources that optimize AI technologies. Aside from giving members more exposure and opportunities, the program helps them grow exponentially, demonstrating Arm’s mission to elevate AI.
Improving Compression Technologies
At the AI Virtual Tech Talk, Nota would introduce ‘NetsPresso,’ an automatic AI model compression technology. Moreover, NetsPresso efficiently compresses intricate AI neural networks and reduces file sizes. As a result, the compressed data turn into lightweight models that take up less disk space and accelerate file transfer between systems.
Compared to current expensive, time-consuming conventional technologies, the NetsPresso platform conducts 100% automatic compression, creating lightweight models rapidly. NetsPresso also removes the need for professional engineering services, significantly cutting down costs while speeding up deep learning inference.
Furthermore, NetsPresso develops lightweight models optimized for target devices, improving compression rates while generating similar accuracy levels as existing models. After hosting the AI Virtual Tech Talk, Nota plans to conduct NetsPresso pilot tests.
In particular, Nota would release NetsPresso’s trial version to selected applicants. NetsPresso’s pilot test would also conduct input model compression and optimization using numerous methods, improving performance and reducing latency.
“We are giving early access for NetsPresso trial version to users before its official release so that we can make sure that the final release version meets customers’ expectations,” said Myungsu Chae, Nota’s CEO.
Overcoming AI Limitations
Nota develops AI model compression solutions for various sectors, including construction, healthcare, mobility, security, and transportation. In addition to developing AI solutions, Nota removes the limitations in low-end devices, giving them ‘brains’ for improved control systems.
To this end, the startup utilizes its extensive research and development related to core deep learning compression technology. Nota also optimizes its AI model compression technologies to lessen cloud-related resource usage.
Some of Nota’s clients include Amazon, Intel, LG Electronics, Microsoft, Naver, Nvidia, Samsung, and SK Telecom. Previously, Nota raised a Series A funding, successfully acquiring around 10 billion won ($8.8 million), becoming the only Samsung and LG-funded startup.
