As AI is becoming more ubiquitous in many research fields and applications, AI models need to be energy efficient. The training of AI models (such as deep networks) requires extensive computational power, especially when large datasets are used, and multiple runs for prototyping and hyper-parameters tuning are needed.
Read more