Training AI algorithms using large-scale synthetic data
Reference persons GIAN PAOLO CIMELLARO
Description The AI-based approaches for monitoring civil infrastructure require a lot of training data to produce accurate and reliable predictions. However, one of the major bottlenecks in this area is the lack of adequate field monitoring data, which restricts the wide application of these techniques. Moreover, manual annotation of the acquired data is labor-intensive and time consuming. Besides, any inadvertent human error in the annotation process may adversely impact the performance of the trained model.
This research aims at overcoming these limitations by means of synthetic data produced in a simulated environment resorting to predictive engineering simulation of physical failure, due to mechanical short-time loading taking into account pre-stress, e.g., due to aging. In a second step, generative adversarial networks are used to generate a large amount of data in a relatively short time based on the expensive high-resolving computation. The simulated environment also permits automatic annotation of the data, which saves a lot of time and effort which go into the manual data annotation process. The synthetic data will then be used to train AI algorithms covering figures as well as (multi-) time series data.
Deadline 17/11/2023 PROPONI LA TUA CANDIDATURA