Helbling increases efficiency through neuromorphic computing

Helbling experts are driving innovations that are made possible by neuromorphic computing technology. According to a specialist article, the engineering and consulting company is positioning itself as an industry partner in this field. Knowledge and experience in this area has been systematically built up with partners in recent times. In the article, the authors attribute great potential to neuromorphic computing as a key technology - especially for edge devices, i.e. devices with local data processing. This is interesting, for example, in system monitoring for autonomous, permanently operational devices. The medtech sector with medical implants and wearables is also mentioned. The advantages are primarily based on higher energy efficiency and computing latency compared to conventional digital neural networks, according to the report. This would mimic natural neuronal networks in the brains of animals.

The experts go into technology details and also describe important aspects of integration. In general, two things are central to neuromorphic computing: the use of in-memory calculations, known as compute-in-memory, and special data coding, known as sparse spike data encoding. For example, the former solves a bottleneck problem of current computer architectures: data and instructions have to be transferred between the memory and processor with every instruction cycle. Internal memory calculations made it possible to combine processing and memory.

Despite all the advantages, the technology faces hurdles according to the analysis. Among other things, there are only a limited number of providers and competition is strong due to the ongoing development of traditional digital devices. In addition, conservative players were hesitant because they had little experience with the theory and use of neuromorphic devices. "However, a new way of thinking is needed to tackle problems relating to hardware and software," the authors write. ce/yvh