When deploying machine learning estimators in science and engineering (SAE) domains, it is critical to avoid failed estimations that can have disastrous consequences, e.g., in aero engine design. This work focuses on detecting and correcting failed state estimations before adopting them in SAE inverse problems, by utilizing simulations and performance metrics guided by physical laws. We suggest to flag a machine learning estimation when its physical model error exceeds a feasible threshold, and propose a novel approach, GEESE, to correct it through optimization, aiming at delivering both low error and high efficiency. The key designs of GEESE include (1) a hybrid surrogate error model to provide fast error estimations to reduce simulation cost and to enable gradient based backpropagation of error feedback, and (2) two generative models to approximate the probability distributions of the candidate states for simulating the exploitation and exploration behaviours. All three models are constructed as neural networks. GEESE is tested on three real-world SAE inverse problems and compared to a number of state-of-the-art optimization/search approaches. Results show that it fails the least number of times in terms of finding a feasible state correction, and requires physical evaluations less frequently in general.

When deploying machine learning estimators in science and engineering (SAE) domains, it is critical to avoid failed estimations that can have disastrous consequences, e.g., in aero engine design.

When deploying machine learning estimators in science and engineering (SAE) domains, it is critical to avoid failed estimations that can have disastrous consequences, e.g., in aero engine design.

When deploying machine learning estimators in science and engineering (SAE) domains, it is critical to avoid failed estimations that can have disastrous consequences, e.g., in aero engine design. This work focuses on detecting and correcting failed state estimations before adopting them in SAE inverse problems, by utilizing simulations and performance metrics guided by physical laws. We suggest to flag a machine learning estimation when its physical model error exceeds a feasible threshold, and propose a novel approach, GEESE, to correct it through optimization, aiming at delivering both low error and high efficiency. The key designs of GEESE include (1) a hybrid surrogate error model to provide fast error estimations to reduce simulation cost and to enable gradient based backpropagation of error feedback, and (2) two generative models to approximate the probability distributions of the candidate states for simulating the exploitation and exploration behaviours. All three models are constructed as neural networks.

Kara Frederick, tech director at the Heritage Foundation, discusses the need for regulations on artificial intelligence as lawmakers and tech titans discuss the potential risks. Artificial intelligence is gaining new uses at a seemingly rapid pace, and the technology is now extending to applications in the animal kingdom. A new AI tool has been developed that allows researchers to use facial recognition to track the faces of geese, which biologists hope will help them understand their way of life and behaviors, according to a report from NPR. Sonia Kleindorfer, a biologist at the University of Vienna, has long studied the species, telling NPR she built off the work of Konrad Lorenz, who studied the birds and was even able to identify individuals within a flock after studying their faces. WHAT IS ARTIFICIAL INTELLIGENCE (AI)? But AI technology could now make that easier, Kleindorfer said, creating databases and using facial recognition to track the movements of a goose with 97% accuracy.

When deploying machine learning estimators in science and engineering (SAE) domains, it is critical to avoid failed estimations that can have disastrous consequences, e.g., in aero engine design. This work focuses on detecting and correcting failed state estimations before adopting them in SAE inverse problems, by utilizing simulations and performance metrics guided by physical laws. We suggest to flag a machine learning estimation when its physical model error exceeds a feasible threshold, and propose a novel approach, GEESE, to correct it through optimization, aiming at delivering both low error and high efficiency. The key designs of GEESE include (1) a hybrid surrogate error model to provide fast error estimations to reduce simulation cost and to enable gradient based backpropagation of error feedback, and (2) two generative models to approximate the probability distributions of the candidate states for simulating the exploitation and exploration behaviours. All three models are constructed as neural networks. GEESE is tested on three real-world SAE inverse problems and compared to a number of state-of-the-art optimization/search approaches. Results show that it fails the least number of times in terms of finding a feasible state correction, and requires physical evaluations less frequently in general.

However, according to the resolution, "humans must always be ultimately responsible for, and able to overrule, decisions" that are taken by new technologies, especially in medical, legal and accounting professions. For the banking sector, the committee calls for a regulatory framework that ensures independent supervision of automated decision-making systems by qualified professionals in cases where the public interest is at stake. This framework should also make it possible for consumers to seek human review when mistakes appear as a result of using this type of new technologies. Likewise, automated decision-making systems should only use high-quality and unbiased data sets and "explainable and unbiased algorithms" to guarantee trust and acceptance, the resolution states. "We have to make sure that consumer protection and trust is ensured and that the data sets used in automated decision-making systems are of high-quality and are unbiased," said Belgian MEP Petra De Sutter (Greens/EFA), who chairs the IMCO committee.