Abstract:The error propagation among estimated parameters reflects the correlation among the parameters. We study the capability of machine learning of "learning" the correlation of estimated parameters. We show that machine learning can recover the relation between the uncertainties of different parameters, especially, as predicted by the error propagation formula. Gravitational lensing can be used to probe both astrophysics and cosmology. As a practical application, we show that the machine learning is able to intelligently find the error propagation among the gravitational lens parameters (effective lens mass $M_{L}$ and Einstein radius $\theta_{E}$) in accordance with the theoretical formula for the singular isothermal ellipse (SIE) lens model. The relation of errors of lens mass and Einstein radius, (e.g. the ratio of standard deviations $\mathcal{F}=\sigma_{\hat{ M_{L}}}/ \sigma_{\hat{\theta_{E}}}$) predicted by the deep convolution neural network are consistent with the error propagation formula of SIE lens model. As a proof-of-principle test, a toy model of linear relation with Gaussian noise is presented. We found that the predictions obtained by machine learning indeed indicate the information about the law of error propagation and the distribution of noise. Error propagation plays a crucial role in identifying the physical relation among parameters, rather than a coincidence relation, therefore we anticipate our case study on the error propagation of machine learning predictions could extend to other physical systems on searching the correlation among parameters.

What problem does this paper attempt to address?

### What problems does this paper attempt to solve? This paper aims to explore the error propagation mechanism when predicting parameters in strong gravitational lensing systems in deep learning. Specifically, the researchers hope to understand whether machine - learning models can capture the correlations between different estimated parameters and verify whether these models can correctly reflect the error relationships between parameters according to the error propagation formula. #### Main problems and goals: 1. **Error propagation and parameter correlation**: The researchers hope to explore whether these errors reflect the intrinsic correlations between parameters by analyzing the prediction of parameter errors by deep neural networks (DNN). 2. **Application in strong gravitational lensing models**: Of particular concern is whether machine - learning models can accurately predict the error relationship between the effective lensing mass $M_L$ and the Einstein radius $\theta_E$, and whether this relationship conforms to the theoretical error propagation formula. 3. **Verification of the linear toy model**: To verify the above assumptions, the researchers designed a simple linear - relationship toy model and introduced Gaussian noise to test the performance of machine - learning models at different noise levels. 4. **Verification in practical applications**: By using convolutional neural networks (CNN) such as VGG16 to model the images of strong gravitational lensing systems, further verify the performance of machine - learning models in actual physical systems. #### Key formulas: - **Error propagation formula**: For two parameters $x$ and $y$, the error propagation formula for their standard deviation $\sigma$ is: \[ \sigma^2(y)=\left(\frac{\partial y}{\partial x}\right)^2\sigma^2(x) \] - **Relationship between effective lensing mass and Einstein radius**: \[ M_L = \frac{c^2D}{4G}\theta_E^2 \] where $D$ is the angular diameter distance, $c$ is the speed of light, and $G$ is the gravitational constant. #### Conclusion: The research results show that deep - learning models can indeed capture the error propagation relationships between parameters to a certain extent, especially in the presence of noise, the performance of the model is consistent with the theoretical error propagation formula. This finding not only helps to understand the working mechanism of deep - learning models but also provides new ideas for finding parameter correlations in other physical systems in the future.

Understanding the predication mechanism of deep learning through error propagation among parameters in strong lensing case

Neural network prediction of model parameters for strong lensing samples from Hyper Suprime-Cam Survey

On machine learning search for gravitational lenses

Modeling lens potentials with continuous neural fields in galaxy-scale strong lenses

Optimizing machine learning methods to discover strong gravitational lenses in the Deep Lens Survey

Forklens: Accurate weak-lensing shear measurement with deep learning

Deep learning method in testing the cosmic distance duality relation

Full-sky Ray-tracing Simulation of Weak Lensing Using ELUCID Simulations: Exploring Galaxy Intrinsic Alignment and Cosmic Shear Correlations

Lensing reconstruction from the cosmic microwave background polarization with machine learning

Neural Network Prediction of Strong Lensing Systems with Domain Adaptation and Uncertainty Quantification

Cosmological model discrimination with Deep Learning

Pixelated Reconstruction of Gravitational Lenses using Recurrent Inference Machines

Lens Modeling of STRIDES Strongly Lensed Quasars using Neural Posterior Estimation

DeepGraviLens: a Multi-Modal Architecture for Classifying Gravitational Lensing Data

Accelerating galaxy dynamical modeling using a neural network for joint lensing and kinematics analyses

Measuring the Substructure Mass Power Spectrum of 23 SLACS Strong Galaxy-Galaxy Lenses with Convolutional Neural Networks

Simple lessons from complex learning: what a neural network model learns about cosmic structure formation

Weak-lensing shear measurement with machine learning: teaching artificial neural networks about feature noise

Constraining the spatial curvature of the local Universe with deep learning

lemon : LEns MOdelling with Neural networks – I. Automated modelling of strong gravitational lenses with Bayesian Neural Networks