NTT launches physics of AI group and AI inference chip design for 4K video

NTT research was announced in the annual increase in the new AI main research group Physics of Artificial Intelligence Group.

Physical AI turned into a big thing in 2025 Conducts the Charge of NVIDIA They can go to the market faster to create synthetic data to create self-driving machines and humanoid robotos. NTT research begins group (PAI) Group (PAI) group (PAI) to go to the board.

The new Independent Group of NTT research closes the exploration (PHI) laboratory to develop a better confidence and security results of AI’s “black box”. NTT research, which has $ 3.6 billion in the budget of $ 3.6 billion, is a department of NTT and Japanese.

Last year, NTT first appeared in the Harvard University’s Brain Science Center, the main contributions made in the last five years and the main contributions of academic partners.

The new group will be followed by Hyenori Tanaka, NTT research scientist and physics, neurology and machine learning specialist, human / EU cooperation.

The new group will continue to develop an interdisciplinary approach to understand Pioneer AI by the team in the last five years.

Early, PHI Laboratory, acutely improved energy efficiency, the importance of understanding the nature of the “black box” of novel systems with energy efficiency. Now a surprising proportion with the EU, reliability and security issues have become critical for the AI’s acceptance industry applications and management.

In cooperation with the leading academic researchers, solving the similarities between the physics of artificial intelligence, biological and artificial intellectuals, further open the complexity of AI mechanisms, and lead to a more harmonious combination of human and AI cooperation. The goal is to get a better understanding of how we can develop a study, knowledge of knowledge and decision-making, future unity, safe and reliable AI.

This approach reflects what physics do in the few centuries: when people apply forces, physics that reveals the exact details of the mechanisms we know today. For example, the development of the steam engine, which in turn, allowed to create advanced semiconductors. Similarly, the case of this group will form the future of AI technology.

New Group Harvard University Brain Science (CBS), Harvard Prof. Stanford University Associate Professor Associate Professor Surya Ganguli, who co-authored Tanaka, plans to cooperate with Surya Ganguli. The main team of the group includes Tanaka, NTT research scientist Maya Okawa and NTT’s doctoral companion Ekdeep Singh Lubana.

The contributions given to the previous date include:

• Wide Neuron Network Pruning Algorithm (more than 750 quotes in total 4 years)
• An unbiased algorithm for a large language models (LLM) recognized by the US National Standards and Technology Institute (NIST) for scientific and practical concepts; and
• New ideas to the dynamics of learning the concepts of AI

The physics of the group of artificial intelligence has three gear mission. 1) intends to deepen my understanding of the EU mechanisms, gentle regulation (ie learning) is better than accessing ethics than a patch work. 2) Debt from experienced physique will continue to create EU’s systemic manageable sites and observe the study and forecasting behavior of AI step by step. 3) Structured operations and information controls to prevent the violation of trust between AI and human operators.

“Today, a new step in the EU’s EU’s concept of EU’s EU’s EU’s concept of EU’s EU,” NTT Research President and CEO Kazu Gomi said today. “The emergence of AI solutions in all areas of the EU and a deep impact on our relations with technology.

The physics of the artificial intelligence group covers an intermediate -bellinjine approach with AI, physics, neurology and psychology. This approach seems to be an ordinary criterion in need of supporting targets such as justice and security that causes a sustainable AI adoption. In terms of energy efficiency, other groups in the PHI Laboratory are already working to reduce energy consumption of Energy computing platforms through optical calculation and road cutting, thin film Lithium Niobate (TFLN) technology. On top of that, inspired by the wide differentiament between the watts consumed by the LLS and a human or animal brain, the new group will also investigate the way they are similar to biological and artificial neuron networks.

“The EU’s existence is harmoniously, we approach the design and implementation of reliability and EI solutions,” he said. “We have a way to explain how the brain concerns and deep learning models with the arising of this group. Beoke, research, physics, neurology and machine learning understanding to bring more natural intelligent algorithms and hardware.”

Since 2019, the PHI Laboratory is conducting research for new ways of calculating systems using photonic-based technologies. TFLN-based devices are investigated with this effort, the arbitrary Ising Machine provides new perspectives on complex optimization problems to solve classic computers.

In addition to the joint research agreement with Harvard, these years have worked over the years along with the California University, Harvard University, Harvard University, Harvard University, Harvard University, Harvard University, University of Technology, Michigan University and NASA Ames Research. In general, the Phi laboratory gave more than 150 documents, five in nature, in a science and twenty natural sister magazines.

NTT announces AI’s result chip for real-time 4k video processing

NTT Corp., at the same time, high-resolution and highest resolution in 4K (FPS) and the highest number of videos in seconds have announced a new, large-scale integration (LSI) for the results of real-time AI. This low power technology is designed for remote terminal accommodation of the edge and power that requires an ultra high-precision video for real-time video processing.

For example, when this LSI is installed a Drone, Adrot, Japan can discover those who have the highest level AI’s highest level AI, 492 feet, or objects. In case of use, the operator includes an expanded Apron-based infrastructure inspection for reducing operations outside the visual line of the operator, reducing labor and expenses.

“The combination of low power AI with ultra high-precision video catches a very large
Gomi also said the amount of potential for infrastructure inspection from infrastructure in public safety, “he said.

EDGE and powerful terminals are limited to power consumption that is lower than the GPU used in AI servers; Watt ten previously compared to hundreds of watts by the second. LSI repels these restrictions by implementing the result of the EI’s AI’s EI. This engine reduces the complexity of computing, when the engine, interface correlation and dynamic bit accuracy, enhances calculation efficiency, improving calculation efficiency. Using this LSI, it is possible only once (Yolov3), with less than 20 watts.

NTT, NTT Innovative Devices plans to recognize this LSI in 2025 through the 2025 financial year through the Corporation Operating Company. NTT has improved this LSI and raised the company’s annual research and innovation summit. On April 9-10, 2025, 2025 are improved in San Francisco.

When looking upon the upcoming researchers learn the application of the Innovative Optical and Wireless Network (IAK) initiative, led by the NTT and Iown Global Forum of the IAK Global Forum. DCI uses high-speed and low level opportunities of the iAnd comprehensive network to solve the problems of modern network infrastructure, including difficulty, performance and restrictions in high energy consumption.

In addition, NTT researchers cooperate with NTT Data, Inc., this LSI’s ownership attribute is cooperating on the progress of this LSI in connection with the Technology Technologies. ABE provides a delicate entry control and flexible policy settings with shared secret encryption technologies that allow reliable data exchange to existing applications and data stores.

Iown ID

Yesterday, NTT, NTT President and CEO Akkira Schimada, and Kawazoe, NTT General President and CTO CTO announced, Iown IDNTT discussed the initiative of the initiative of the Global Inthto Intht (Innovative Optical and Wireless Network)
The leader of the technology.

The newly translated book will allow NTT to see how Iown’s sight and more steadfast society.

Now the NTT annual research and innovation summit and the Innovation Summit are already available in the “iowown ID” in Amazon. On April 9-10, 2025, 2025 are improved in San Francisco.