Important results from the project
We have implemented active learning for ML models used to design microwave circuits in an internal cloud environment at Ericsson and achieved expected performance as active learning was shown to result in better ML models for the same amount of training data.
Expected long term effects
That the development of certain types of ML models becomes more efficient by applying active learning.
Approach and implementation
The project began with the student reviewing relevant scientific literature and familiarizing himself with what we at Ericsson have done so far in ML-assisted circuit design. Then, an ML model suitable for active learning was developed and a learning loop was implemented in a cloud environment at Ericsson. Active learning was then compared to conventional random data generation and was shown to have significant advantages.
The project description has been provided by the project members themselves and the text has not been looked at by our editors.
