Japan Patent Office (JPO) has issued a patent on the artificial intelligence-based image recognition technology by FPT Software, a subsidiary of FPT Corporation. The patent is a significant embodiment of the company’s strength and excellence in world-class research, as well as representing FPT Software’s commitment to deliver AI-enabled services and solutions to hundreds of clients globally.
The methodology, data processing mechanism and AI architecture is protected by JPO across Japan territory, which means either partial or full adoption of the technology into any product development by an individual or business, will need to be licensed by FPT Software.
The patent granted by JPO for FPT Software’s AI technology in May 2023
This technology introduces a new neuron mapping architecture that de-aggregates imaging components while still retaining the spatial connection of data, across molecular spots on the surface of the image. This neuron mapping architecture is not only applicable in local-density approximation exercises (examples include for people or vehicles in traffic), but also for other applications in computer vision.
The inventor, Dr. Dao Huu Hung, is currently the Principal AI Scientist at AI Center, FPT Software. According to Dr. Hung, the data processing mechanism introduced in this patent is similar to that of Vision Transformer (ViT) models, the variation of well-known Transformer models in ChatGPT (Generative Pretrained Transformer) in the field of image recognition.
“FPT Software has filed the application for the patent in August 2019, one year prior to ViT, which was firstly released in October 2020. As a result, we have submitted an extended application to expand the protection scope of our technology, which covers a part of ViT’s”, Dr. Hung elaborated. “If successful, we will continue filing to seek protection in the US and Europe, where ViT technology has been integrated into many commercial products”.
With Mila’s consultation, FPT Software will also establish its framework for Artificial Intelligence ethics to ensure reliability, safety, transparency and fairness in the research, development and application of AI technology.