arXiv subjects

Michael Eickenberg

Publications and source records attributed to Michael Eickenberg.

At least 37 records · Page 2Linked to original sources

arXiv2023

Learnable wavelet neural networks for cosmological inference

Convolutional neural networks (CNNs) have been shown to both extract more information than the traditional two-point statistics from cosmological fields, and marginalise over astrophysical effects extremely well. However, CNNs require large amounts of training data, which is potentially problematic in the domain of expensive cosmological simulations, and it is difficult to interpret the network. In this work we apply the learnable scattering transform, a kind of convolutional neural network that uses trainable wavelets as filters, to the problem of cosmological inference and marginalisation over astrophysical effects. We present two models based on the scattering transform, one constructed for performance, and one constructed for interpretability, and perform a comparison with a CNN. We find that scattering architectures are able to outperform a CNN, significantly in the case of small training data samples. Additionally we present a lightweight scattering network that is highly interpretable.

astro-ph.IM↗

arXiv2023

Can Forward Gradient Match Backpropagation?

Forward Gradients - the idea of using directional derivatives in forward differentiation mode - have recently been shown to be utilizable for neural network training while avoiding problems generally associated with backpropagation gradient computation, such as locking and memorization requirements. The cost is the requirement to guess the step direction, which is hard in high dimensions. While current solutions rely on weighted averages over isotropic guess vector distributions, we propose to strongly bias our gradient guesses in directions that are much more promising, such as feedback obtained from small, local auxiliary networks. For a standard computer vision neural network, we conduct a rigorous study systematically covering a variety of combinations of gradient targets and gradient guesses, including those previously presented in the literature. We find that using gradients obtained from a local loss as a candidate direction drastically improves on random noise in Forward Gradient methods.

cs.LG↗

arXiv2023

Adversarial Attacks on the Interpretation of Neuron Activation Maximization

The internal functional behavior of trained Deep Neural Networks is notoriously difficult to interpret. Activation-maximization approaches are one set of techniques used to interpret and analyze trained deep-learning models. These consist in finding inputs that maximally activate a given neuron or feature map. These inputs can be selected from a data set or obtained by optimization. However, interpretability methods may be subject to being deceived. In this work, we consider the concept of an adversary manipulating a model for the purpose of deceiving the interpretation. We propose an optimization framework for performing this manipulation and demonstrate a number of ways that popular activation-maximization interpretation techniques associated with CNNs can be manipulated to change the interpretations, shedding light on the reliability of these methods.

cs.LG↗

arXiv2023

Robust Simulation-Based Inference in Cosmology with Bayesian Neural Networks

Simulation-based inference (SBI) is rapidly establishing itself as a standard machine learning technique for analyzing data in cosmological surveys. Despite continual improvements to the quality of density estimation by learned models, applications of such techniques to real data are entirely reliant on the generalization power of neural networks far outside the training distribution, which is mostly unconstrained. Due to the imperfections in scientist-created simulations, and the large computational expense of generating all possible parameter combinations, SBI methods in cosmology are vulnerable to such generalization issues. Here, we discuss the effects of both issues, and show how using a Bayesian neural network framework for training SBI can mitigate biases, and result in more reliable inference outside the training set. We introduce cosmoSWAG, the first application of Stochastic Weight Averaging to cosmology, and apply it to SBI trained for inference on the cosmic microwave background.

astro-ph.CO↗

arXiv2023

Local Learning with Neuron Groups

Traditional deep network training methods optimize a monolithic objective function jointly for all the components. This can lead to various inefficiencies in terms of potential parallelization. Local learning is an approach to model-parallelism that removes the standard end-to-end learning setup and utilizes local objective functions to permit parallel learning amongst model components in a deep network. Recent works have demonstrated that variants of local learning can lead to efficient training of modern deep networks. However, in terms of how much computation can be distributed, these approaches are typically limited by the number of layers in a network. In this work we propose to study how local learning can be applied at the level of splitting layers or modules into sub-components, adding a notion of width-wise modularity to the existing depth-wise modularity associated with local learning. We investigate local-learning penalties that permit such models to be trained efficiently. Our experiments on the CIFAR-10, CIFAR-100, and Imagenet32 datasets demonstrate that introducing width-level modularity can lead to computational advantages over existing methods based on local learning and opens new opportunities for improved model-parallel distributed training. Code is available at: https://github.com/adeetyapatel12/GN-DGL.

cs.LG↗

arXiv2022

${\rm S{\scriptsize IM}BIG}$: A Forward Modeling Approach To Analyzing Galaxy Clustering

We present the first-ever cosmological constraints from a simulation-based inference (SBI) analysis of galaxy clustering from the new ${\rm S{\scriptsize IM}BIG}$ forward modeling framework. ${\rm S{\scriptsize IM}BIG}$ leverages the predictive power of high-fidelity simulations and provides an inference framework that can extract cosmological information on small non-linear scales, inaccessible with standard analyses. In this work, we apply ${\rm S{\scriptsize IM}BIG}$ to the BOSS CMASS galaxy sample and analyze the power spectrum, $P_\ell(k)$, to $k_{\rm max}=0.5\,h/{\rm Mpc}$. We construct 20,000 simulated galaxy samples using our forward model, which is based on high-resolution ${\rm Q{\scriptsize UIJOTE}}$ $N$-body simulations and includes detailed survey realism for a more complete treatment of observational systematics. We then conduct SBI by training normalizing flows using the simulated samples and infer the posterior distribution of $Λ$CDM cosmological parameters: $Ω_m, Ω_b, h, n_s, σ_8$. We derive significant constraints on $Ω_m$ and $σ_8$, which are consistent with previous works. Our constraints on $σ_8$ are $27\%$ more precise than standard analyses. This improvement is equivalent to the statistical gain expected from analyzing a galaxy sample that is $\sim60\%$ larger than CMASS with standard methods. It results from additional cosmological information on non-linear scales beyond the limit of current analytic models, $k > 0.25\,h/{\rm Mpc}$. While we focus on $P_\ell$ in this work for validation and comparison to the literature, ${\rm S{\scriptsize IM}BIG}$ provides a framework for analyzing galaxy clustering using any summary statistic. We expect further improvements on cosmological constraints from subsequent ${\rm S{\scriptsize IM}BIG}$ analyses of summary statistics beyond $P_\ell$.

astro-ph.CO↗

arXiv2022

${\rm S{\scriptsize IM}BIG}$: Mock Challenge for a Forward Modeling Approach to Galaxy Clustering

Simulation-Based Inference of Galaxies (${\rm S{\scriptsize IM}BIG}$) is a forward modeling framework for analyzing galaxy clustering using simulation-based inference. In this work, we present the ${\rm S{\scriptsize IM}BIG}$ forward model, which is designed to match the observed SDSS-III BOSS CMASS galaxy sample. The forward model is based on high-resolution ${\rm Q{\scriptsize UIJOTE}}$ $N$-body simulations and a flexible halo occupation model. It includes full survey realism and models observational systematics such as angular masking and fiber collisions. We present the "mock challenge" for validating the accuracy of posteriors inferred from ${\rm S{\scriptsize IM}BIG}$ using a suite of 1,500 test simulations constructed using forward models with a different $N$-body simulation, halo finder, and halo occupation prescription. As a demonstration of ${\rm S{\scriptsize IM}BIG}$, we analyze the power spectrum multipoles out to $k_{\rm max} = 0.5\,h/{\rm Mpc}$ and infer the posterior of $Λ$CDM cosmological and halo occupation parameters. Based on the mock challenge, we find that our constraints on $Ω_m$ and $σ_8$ are unbiased, but conservative. Hence, the mock challenge demonstrates that ${\rm S{\scriptsize IM}BIG}$ provides a robust framework for inferring cosmological parameters from galaxy clustering on non-linear scales and a complete framework for handling observational systematics. In subsequent work, we will use ${\rm S{\scriptsize IM}BIG}$ to analyze summary statistics beyond the power spectrum including the bispectrum, marked power spectrum, skew spectrum, wavelet statistics, and field-level statistics.

astro-ph.CO↗

arXiv2022

Towards a non-Gaussian Generative Model of large-scale Reionization Maps

High-dimensional data sets are expected from the next generation of large-scale surveys. These data sets will carry a wealth of information about the early stages of galaxy formation and cosmic reionization. Extracting the maximum amount of information from the these data sets remains a key challenge. Current simulations of cosmic reionization are computationally too expensive to provide enough realizations to enable testing different statistical methods, such as parameter inference. We present a non-Gaussian generative model of reionization maps that is based solely on their summary statistics. We reconstruct large-scale ionization fields (bubble spatial distributions) directly from their power spectra (PS) and Wavelet Phase Harmonics (WPH) coefficients. Using WPH, we show that our model is efficient in generating diverse new examples of large-scale ionization maps from a single realization of a summary statistic. We compare our model with the target ionization maps using the bubble size statistics, and largely find a good agreement. As compared to PS, our results show that WPH provide optimal summary statistics that capture most of information out of a highly non-linear ionization fields.

astro-ph.CO↗

arXiv2022

HIFlow: Generating Diverse HI Maps and Inferring Cosmology while Marginalizing over Astrophysics using Normalizing Flows

A wealth of cosmological and astrophysical information is expected from many ongoing and upcoming large-scale surveys. It is crucial to prepare for these surveys now and develop tools that can efficiently extract most information. We present HIFlow: a fast generative model of the neutral hydrogen (HI) maps that is conditioned only on cosmology ($Ω_{m}$ and $σ_{8}$) and designed using a class of normalizing flow models, the Masked Autoregressive Flow (MAF). HIFlow is trained on the state-of-the-art simulations from the Cosmology and Astrophysics with MachinE Learning Simulations (CAMELS) project. HIFlow has the ability to generate realistic diverse maps without explicitly incorporating the expected 2D maps structure into the flow as an inductive bias. We find that HIFlow is able to reproduce the CAMELS average and standard deviation HI power spectrum (Pk) within a factor of $\lesssim$ 2, scoring a very high $R^{2} > 90\%$. By inverting the flow, HIFlow provides a tractable high-dimensional likelihood for efficient parameter inference. We show that the conditional HIFlow on cosmology is successfully able to marginalize over astrophysics at the field level, regardless of the stellar and AGN feedback strengths. This new tool represents a first step toward a more powerful parameter inference, maximizing the scientific return of future HI surveys, and opening a new avenue to minimize the loss of complex information due to data compression down to summary statistics.

astro-ph.CO↗

arXiv2022

Parametric Scattering Networks

The wavelet scattering transform creates geometric invariants and deformation stability. In multiple signal domains, it has been shown to yield more discriminative representations compared to other non-learned representations and to outperform learned representations in certain tasks, particularly on limited labeled data and highly structured signals. The wavelet filters used in the scattering transform are typically selected to create a tight frame via a parameterized mother wavelet. In this work, we investigate whether this standard wavelet filterbank construction is optimal. Focusing on Morlet wavelets, we propose to learn the scales, orientations, and aspect ratios of the filters to produce problem-specific parameterizations of the scattering transform. We show that our learned versions of the scattering transform yield significant performance gains in small-sample classification settings over the standard scattering transform. Moreover, our empirical results suggest that traditional filterbank constructions may not always be necessary for scattering transforms to extract effective representations.

cs.LG↗

arXiv2022

Generative Models of Multi-channel Data from a Single Example -- Application to Dust Emission

The quest for primordial $B$-modes in the cosmic microwave background has emphasized the need for refined models of the Galactic dust foreground. Here, we aim at building a realistic statistical model of the multi-frequency dust emission from a single example. We introduce a generic methodology relying on microcanonical gradient descent models conditioned by an extended family of wavelet phase harmonic (WPH) statistics. To tackle the multi-channel aspect of the data, we define cross-WPH statistics, quantifying non-Gaussian correlations between maps. Our data-driven methodology could apply to various contexts, and we have updated the software PyWPH, on which this work relies, accordingly. Applying this to dust emission maps built from a magnetohydrodynamics simulation, we construct and assess two generative models of: 1) a $(I, E, B)$ multi-observable input, 2) a $\{I_ν\}_ν$ multi-frequency input. The samples exhibit consistent features compared to the original maps. A statistical analysis of 1) shows that the power spectra, distributions of pixels, and Minkowski functionals are captured to a good extent. We analyze 2) by fitting the spectral energy distribution (SED) of both the synthetic and original maps with a modified blackbody (MBB) law. The maps are equally well fitted, and a comparison of the MBB parameters shows that our model succeeds in capturing the spatial variations of the SED from the data. Besides the perspectives of this work for dust emission modeling, the introduction of cross-WPH statistics opens a new avenue to characterize non-Gaussian interactions across different maps, which we believe will be fruitful for astrophysics.

astro-ph.CO↗

arXiv2022

TNT: Vision Transformer for Turbulence Simulations

Turbulence is notoriously difficult to model due to its multi-scale nature and sensitivity to small perturbations. Classical solvers of turbulence simulation generally operate on finer grids and are computationally inefficient. In this paper, we propose the Turbulence Neural Transformer (TNT), which is a learned simulator based on the transformer architecture, to predict turbulent dynamics on coarsened grids. TNT extends the positional embeddings of vanilla transformers to a spatiotemporal setting to learn the representation in the 3D time-series domain, and applies Temporal Mutual Self-Attention (TMSA), which captures adjacent dependencies, to extract deep and dynamic features. TNT is capable of generating comparatively long-range predictions stably and accurately, and we show that TNT outperforms the state-of-the-art U-net simulator on several metrics. We also test the model performance with different components removed and evaluate robustness to different initial conditions. Although more experiments are needed, we conclude that TNT has great potential to outperform existing solvers and generalize to additional simulation datasets.

physics.flu-dyn↗

arXiv2022

Cosmological Information in the Marked Power Spectrum of the Galaxy Field

Marked power spectra are two-point statistics of a marked field obtained by weighting each location with a function that depends on the local density around that point. We consider marked power spectra of the galaxy field in redshift space that up-weight low density regions, and perform a Fisher matrix analysis to assess the information content of this type of statistics using the Molino mock catalogs built upon the Quijote simulations. We identify four different ways to up-weight the galaxy field, and compare the Fisher information contained in their marked power spectra to the one of the standard galaxy power spectrum, when considering monopole and quadrupole of each statistic. Our results show that each of the four marked power spectra can tighten the standard power spectrum constraints on the cosmological parameters $Ω_{\rm m}$, $Ω_{\rm b}$, $h$, $n_s$, $M_ν$ by $15-25\%$ and on $σ_8$ by a factor of 2. The same analysis performed by combining the standard and four marked power spectra shows a substantial improvement compared to the power spectrum constraints that is equal to a factor of 6 for $σ_8$ and $2.5-3$ for the other parameters. Our constraints may be conservative, since the galaxy number density in the Molino catalogs is much lower than the ones in future galaxy surveys, which will allow them to probe lower density regions of the large-scale structure.

astro-ph.CO↗

arXiv2022

Kymatio: Scattering Transforms in Python

The wavelet scattering transform is an invariant signal representation suitable for many signal processing and machine learning applications. We present the Kymatio software package, an easy-to-use, high-performance Python implementation of the scattering transform in 1D, 2D, and 3D that is compatible with modern deep learning frameworks. All transforms may be executed on a GPU (in addition to CPU), offering a considerable speed up over CPU implementations. The package also has a small memory footprint, resulting inefficient memory usage. The source code, documentation, and examples are available undera BSD license at https://www.kymat.io/

cs.LG↗

arXiv2022

Wavelet Moments for Cosmological Parameter Estimation

Extracting non-Gaussian information from the non-linear regime of structure formation is key to fully exploiting the rich data from upcoming cosmological surveys probing the large-scale structure of the universe. However, due to theoretical and computational complexities, this remains one of the main challenges in analyzing observational data. We present a set of summary statistics for cosmological matter fields based on 3D wavelets to tackle this challenge. These statistics are computed as the spatial average of the complex modulus of the 3D wavelet transform raised to a power $q$ and are therefore known as invariant wavelet moments. The 3D wavelets are constructed to be radially band-limited and separable on a spherical polar grid and come in three types: isotropic, oriented, and harmonic. In the Fisher forecast framework, we evaluate the performance of these summary statistics on matter fields from the Quijote suite, where they are shown to reach state-of-the-art parameter constraints on the base $Λ$CDM parameters, as well as the sum of neutrino masses. We show that we can improve constraints by a factor 5 to 10 in all parameters with respect to the power spectrum baseline.

astro-ph.CO↗

arXiv2022

The CAMELS project: public data release

The Cosmology and Astrophysics with MachinE Learning Simulations (CAMELS) project was developed to combine cosmology with astrophysics through thousands of cosmological hydrodynamic simulations and machine learning. CAMELS contains 4,233 cosmological simulations, 2,049 N-body and 2,184 state-of-the-art hydrodynamic simulations that sample a vast volume in parameter space. In this paper we present the CAMELS public data release, describing the characteristics of the CAMELS simulations and a variety of data products generated from them, including halo, subhalo, galaxy, and void catalogues, power spectra, bispectra, Lyman-$α$ spectra, probability distribution functions, halo radial profiles, and X-rays photon lists. We also release over one thousand catalogues that contain billions of galaxies from CAMELS-SAM: a large collection of N-body simulations that have been combined with the Santa Cruz Semi-Analytic Model. We release all the data, comprising more than 350 terabytes and containing 143,922 snapshots, millions of halos, galaxies and summary statistics. We provide further technical details on how to access, download, read, and process the data at \url{https://camels.readthedocs.io}.

astro-ph.CO↗

arXiv2021

The CAMELS Multifield Dataset: Learning the Universe's Fundamental Parameters with Artificial Intelligence

We present the Cosmology and Astrophysics with MachinE Learning Simulations (CAMELS) Multifield Dataset, CMD, a collection of hundreds of thousands of 2D maps and 3D grids containing many different properties of cosmic gas, dark matter, and stars from 2,000 distinct simulated universes at several cosmic times. The 2D maps and 3D grids represent cosmic regions that span $\sim$100 million light years and have been generated from thousands of state-of-the-art hydrodynamic and gravity-only N-body simulations from the CAMELS project. Designed to train machine learning models, CMD is the largest dataset of its kind containing more than 70 Terabytes of data. In this paper we describe CMD in detail and outline a few of its applications. We focus our attention on one such task, parameter inference, formulating the problems we face as a challenge to the community. We release all data and provide further technical details at https://camels-multifield-dataset.readthedocs.io.

cs.LG↗

arXiv2021

Decoupled Greedy Learning of CNNs for Synchronous and Asynchronous Distributed Learning

A commonly cited inefficiency of neural network training using back-propagation is the update locking problem: each layer must wait for the signal to propagate through the full network before updating. Several alternatives that can alleviate this issue have been proposed. In this context, we consider a simple alternative based on minimal feedback, which we call Decoupled Greedy Learning (DGL). It is based on a classic greedy relaxation of the joint training objective, recently shown to be effective in the context of Convolutional Neural Networks (CNNs) on large-scale image classification. We consider an optimization of this objective that permits us to decouple the layer training, allowing for layers or modules in networks to be trained with a potentially linear parallelization. With the use of a replay buffer we show that this approach can be extended to asynchronous settings, where modules can operate and continue to update with possibly large communication delays. To address bandwidth and memory issues we propose an approach based on online vector quantization. This allows to drastically reduce the communication bandwidth between modules and required memory for replay buffers. We show theoretically and empirically that this approach converges and compare it to the sequential solvers. We demonstrate the effectiveness of DGL against alternative approaches on the CIFAR-10 dataset and on the large-scale ImageNet dataset.

cs.LG↗