Towards Automating the AI Operations Lifecycle
Matthew Arnold, Jeffrey Boston, et al.
MLSys 2020
The ever-increasing complexity of applications and networks management demands alternative solutions for identifying of suitable deployment configurations in line with established Service Level Agreements. Some existing approaches utilize methods such as mathematical simulation and application re-enactment under a simulated load implemented as digital twin. In this paper, we describe a mixed-modal framework for varying composition of the above approaches hence improving the performance of application’s ser-vice level modeling further. This framework facilitates the best simulation model performance at any given time by anticipating the model drift and by continuously monitoring performance of the mixed-modal simulation model.
Matthew Arnold, Jeffrey Boston, et al.
MLSys 2020
Saurabh Pujar, Luca Buratti, et al.
DAC 2023
Shubhi Asthana, Bing Zhang, et al.
INFORMS 2022
Hongyi Bian, Rong N. Chang, et al.
SSE 2023