Samsung introduces its neural network, Gauss, which works with images, text and generates code

This week, Samsung officially unveiled its own generative neural network called Gauss. The announcement took place during the annual Samsung AI Forum, which has been held in Seoul since 2017 and focuses on advancements in the field of artificial intelligence.

The South Korean tech giant named its generative AI algorithm in honor of the renowned mathematician Carl Friedrich Gauss, who formulated the Gaussian distribution law, extensively used in machine learning. Samsung expressed that the name embodies its vision for AI models, which revolves around harnessing all the world's knowledge to enhance AI and improve the quality of life for consumers globally.

The development of the Gauss algorithm was spearheaded by engineers at Samsung Research, the company's research division. Currently, the algorithm is being utilized to enhance productivity among employees within the company. However, Samsung plans to expand its availability, making it accessible in consumer-facing applications in the future.

Samsung Gauss is underpinned by three core models: Gauss Language, Gauss Image, and Gauss Code. Gauss Language is geared towards processing textual queries, improving efficiency across various tasks, including email composition, document summarization, language translation, and more. It comprises multiple language models, making it versatile for use both in the cloud and on devices.

The code.i, a coding assistant, is designed to expedite the software development process. Samsung believes it will enable developers to write code more swiftly, offering an interactive interface with support for functions such as code explanations and creating test examples.

Gauss Image, on the other hand, is dedicated to image creation and editing. It facilitates easy style adjustments, addition of new elements, and enhancement of low-resolution image quality. Samsung envisions these models to play a pivotal role in fostering AI's capabilities and usefulness across various domains.