The collaboration seeks to develop new algorithms and innovative techniques, requiring far less data samples while exceeding the results of current AI models. This will automate and minimize human labor-intensive and time-consuming dataset annotation tasks.
Another research area will be the application of advanced 'reinforcement learning' AI algorithms to automatically adjust network parameters by learning and observing real-time performance. This will maximize user experience, especially in dynamic high-traffic scenarios, and simplify network operations.
Mikael Eriksson, Head of Ericsson Japan, says: "We have been exploring how emerging technologies in the Machine Intelligence area can contribute to improvement of network design, operation and performance. We hope that this collaboration will result in new algorithms for applying machine intelligence to LTE and, eventually, 5G networks of our operator customers."
Dr. Kohei Shiomoto, Professor of Tokyo City University, says: "Our laboratory has been focusing on data-driven management that applies data mining technologies to address the increasing complexity of network systems. We hope that our research outcomes from this collaboration are applied to solve the real-world problems in the industry."