Modern technological advancements continue to push the boundaries of what is possible, particularly in the realm of data analysis and AI-driven solutions. One recent breakthrough is NTT's development of Large Language Model (LLM)-based visual machine reading technology. It's an innovation that addresses the challenges of rapidly computing similarity and correspondence between large-scale data sets with high speed and accuracy, making possible more efficient generative AI and media processing.

The core issue NTT's technology aims to solve is the computationally intensive nature of the optimal transport problem. Optimal transport is a mathematical concept that involves finding the most efficient way to move data points from one distribution to another while minimizing cost. The problem is traditionally resource-heavy and time-consuming, making it impractical for large-scale data applications.

NTT has developed an algorithm that takes advantage of cyclic symmetry, a property where a structure remains unchanged under transformations like rotation or inversion—such as the regular patterns seen in gears or snowflakes, for example. By exploiting the symmetry inherent in real-world data, the algorithm breaks down the optimal transport problem into smaller, more manageable components. Doing so transforms the original large-scale problem into a smaller optimization task, with significantly fewer variables. In that way, the algorithm can solve the problem faster and more efficiently than traditional methods.

What does this mean in the real world?

Far-reaching implications. One potential application is in the enhancement and transmittance of visual cognitive abilities. For instance, athletes could use the technology to improve their performance by comparing their movements to those of top performers. Surgeons might refine their techniques by visualizing and analyzing the micro-movements of their peers, leading to better precision and outcomes in complex procedures. Additionally, the technology could be used to create immersive experiences, offering more realistic and engaging experiences for users.

NTT presented these groundbreaking results at the 38th Annual AAAI Conference on Artificial Intelligence, underscoring the significance of their research in the AI community. This forum provided a platform to showcase how their algorithm not only theoretically but also experimentally outperforms traditional methods in solving optimal transport problems.

Looking ahead, NTT plans to continue refining the technology to enhance visual and cognitive abilities across a wide range of applications. As a foundational component of the Innovative Optical and Wireless Network (IOWN), visual machine reading technology is poised to connect people worldwide, fostering collaboration and expanding human capabilities. NTT's development of a fast and efficient algorithm for optimal transport is a leap forward in data analysis and AI-driven solutions. It simplifies complex problems and makes possible rapid and accurate computations. With its potential to bridge the gap between experts and novices, it holds the promise of democratizing access to high-level skills and knowledge.

Ultimately, the technology aims to give NTT's own "tsuzumi" LLM human-like vision and lead to the realization of AI that understands layout and visual elements such as diagrams, graphs and icons, creating new value in collaboration with humans. It's a bright future, blending the unique abilities of humans with the number-crunching power of computers.

NTT—Innovating the Future of AI

Daniel O'Connor

Daniel O'Connor joined the NTT Group in 1999 when he began work as the Public Relations Manager of NTT Europe. While in London, he liaised with the local press, created the company's intranet site, wrote technical copy for industry magazines and managed exhibition stands from initial design to finished displays.

Later seconded to the headquarters of NTT Communications in Tokyo, he contributed to the company's first-ever winning of global telecoms awards and the digitalisation of internal company information exchange.

Since 2015 Daniel has created content for the Group's Global Leadership Institute, the One NTT Network and is currently working with NTT R&D teams to grow public understanding of the cutting-edge research undertaken by the NTT Group.