New Developments in Machine Learning Part 7 (November 2022 Edition) | by Monodeep Mukherjee | November 2022
- Towards efficient and generic entanglement detection by machine learning(arXiv)
Author : Jue Xu, Qi Zhao
Summary : Detecting entanglement is an indispensable step in practical quantum computing and communication. Compared to the conventional method of fidelity-based entanglement witnesses, we offer a flexible approach, machine learningassisted entanglement detection protocol that is robust to different types of noises and sample efficient. In this protocol, an entanglement classifier for a generic entangled state is obtained by forming a classical machine learning model with a synthetic dataset. The data set contains classical characteristics of two types of states and their labels (entangled or separable). The classical characteristics of a state, which are the expectation values of a set of k-local Pauli observables, are efficiently estimated by the classical shadow method. In the numerical simulation, our classifier can detect the entanglement of 4-qubit GHZ states with coherent noise and W states mixed with large white noise, with high accuracy.
2. Fit formulas for learning steady-state causal models from closed-loop operational data(arXiv)
Author : Kristian Løvland, Bjarne Grimstad, Lars Struen Imsland
Summary : The steady-state models that have been learned from historical operational data may be unsuitable for model-based optimization unless the correlations in the training data that are introduced by the control are taken into account. Using recent results from work on structural dynamic causal models, we derive a formula to adjust for this confounding of control, allowing the estimation of a steady-state causal model from looping steady-state data. closed. The formula assumes that the available data were collected under a fixed control law. It works by estimating and taking into account the disturbance that the controller seeks to counter, and allows learning from data collected both by anticipation and by feedback.
3.Debiasing methods for more accurate neural models in vision and language research: a survey (arXiv)
Author : Otavio Parraga, Martin D. More, Christian M. Oliveira, Nathan S. Gavenski, Lucas S. Kupssinskü, Adilson Medronha, Luis V. Moura, Gabriel S. Simões, Rodrigo C. Barros
Summary : Despite being responsible for leading-edge results in several computer vision and natural language processing tasks, neural networks have come under heavy criticism due to some of their current shortcomings. One of them is that neural networks are the correlation machinery prone to model biases in the data instead of focusing on actual useful causal relationships. This issue is particularly severe in application domains affected by aspects such as race, gender, and age. To prevent models from involving unfair decision-making, the AI community has focused its efforts on correcting for algorithmic biases, giving rise to the area of research now widely known as fairness in AI. . In this investigative article, we provide an in-depth overview of the main debiasing methods for fairness-aware neural networks in the context of vision and language research. We propose a new taxonomy to better organize the literature on methods of debiasing for equity, and we discuss current challenges, trends, and important future work directions for the interested researcher and practitioner.