Revolution in computer technology: Greifswald research inspired by the brain!

Revolution in computer technology: Greifswald research inspired by the brain!

Physicists at the University of Greifswald are developing a promising approach for energy-efficient computers that is inspired by the human brain. Given the challenges facing today's computing technology - particularly high energy consumption, the separation of storage and processing units, and slow data transfers - a rethinking of computing architecture is necessary. Growing requirements due to extensive AI models and immense amounts of data are driving research into neuromorphic concepts that are based on the function of the human brain. These approaches are becoming increasingly relevant to achieve sustainable development in computer science, such as uni-greifswald.de reported.

The research team led by Dr. Tahereh Sadat Parvini and Prof. Dr. Markus Münzenberg is working on magnetic tunnel junctions (MTJs) that can both store and process information. The team is developing a hybrid optoelectric excitation scheme that combines electrical currents with short laser pulses. This methodology allows the generation of large thermoelectric voltages in MTJs, which promote synapse-like behavior.

Properties and applications of the new technology

The magnetic tunnel contacts are characterized by three remarkable properties: First, the voltage can be flexibly adjusted, which corresponds to a synaptic weight. Second, spontaneous “spike” signals occur, similar to the exchange of information between nerve cells. Third, a developed neuromorphic network achieved 93.7% recognition accuracy for handwritten digits in simulations. Prof. Dr. Markus Münzenberg highlights the compact and energy-saving platform, which makes this technology predestined for future computing applications. In addition, the technology is compatible with existing semiconductor technology, which makes it possible to use it in everyday devices and high-performance computers.

The challenges in the current computing area, such as the increasing costs of chip development and production and the necessary focus on resource-saving technologies, are also addressed by iis.fraunhofer.de thematised. Neuromorphic computing is seen as a solution because it mimics the way the biological brain works. These approaches not only improve energy efficiency, but also enable resource-intensive AI applications on battery-powered devices.

Future prospects in neuromorphic computing

A key aspect of neuromorphic computing is the combination of low latency and high energy efficiency, which is intended to help optimize real-time edge AI applications. This technology could play a key role, particularly in the area of ​​data protection solutions that do not require access to cloud systems. Fraunhofer IIS has therefore initiated the “Neuromorphic Computing” project, which develops algorithms and hardware for neuromorphic processors in CMOS technology for integration into end devices.

In addition, the industry is working on the development of innovative edge AI applications that will enable high parallel processing and low latency. Major companies such as Intel, IBM and other research institutions are investing heavily in this technology, which could be used in the medium term in areas such as robotics, medical technology and autonomous systems techzeitgeist.de predicted.

Although there are developments such as Intel's Loihi chip, which is specifically optimized for edge computing applications, challenges such as high production costs and the need to develop appropriate software stand in the way of the spread of neuromorphic systems. Forecasts show that the first neuromorphic chips could be available by 2025, but they cannot be easily integrated into the mass market due to existing hurdles.

The developments in neuromorphic computing technology, driven by the University of Greifswald and supported by collaborations with institutes such as the Max Planck Institute for the Science of Light and the International Iberian Nanotechnology Laboratory, represent a significant advance that could not only revolutionize computer science, but will also have an impact on many other industries.