The formation of aerosol particles in the atmosphere impacts air quality and climate change, but many of the organic molecules involved remain unknown. Machine learning could aid in identifying these compounds through accelerated analysis of molecular properties and detection characteristics. However, such progress is hindered by the current lack of curated datasets for atmospheric molecules and their associated properties. To tackle this challenge, we propose a similarity analysis that connects atmospheric compounds to existing large molecular datasets used for machine learning development. We find a small overlap between atmospheric and non-atmospheric molecules using standard molecular representations in machine learning applications. The identified out-of-domain character of atmospheric compounds is related to their distinct functional groups and atomic composition. Our investigation underscores the need for collaborative efforts to gather and share more molecular-level atmospheric chemistry data. The presented similarity based analysis can be used for future dataset curation for machine learning development in the atmospheric sciences.
{"title":"Similarity-Based Analysis of Atmospheric Organic Compounds for Machine Learning Applications","authors":"Hilda Sandström, Patrick Rinke","doi":"arxiv-2406.18171","DOIUrl":"https://doi.org/arxiv-2406.18171","url":null,"abstract":"The formation of aerosol particles in the atmosphere impacts air quality and\u0000climate change, but many of the organic molecules involved remain unknown.\u0000Machine learning could aid in identifying these compounds through accelerated\u0000analysis of molecular properties and detection characteristics. However, such\u0000progress is hindered by the current lack of curated datasets for atmospheric\u0000molecules and their associated properties. To tackle this challenge, we propose\u0000a similarity analysis that connects atmospheric compounds to existing large\u0000molecular datasets used for machine learning development. We find a small\u0000overlap between atmospheric and non-atmospheric molecules using standard\u0000molecular representations in machine learning applications. The identified\u0000out-of-domain character of atmospheric compounds is related to their distinct\u0000functional groups and atomic composition. Our investigation underscores the\u0000need for collaborative efforts to gather and share more molecular-level\u0000atmospheric chemistry data. The presented similarity based analysis can be used\u0000for future dataset curation for machine learning development in the atmospheric\u0000sciences.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141500971","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We have studied the fluctuation (noise) in the position of sliding blocks under constant driving forces on different substrate surfaces. The experimental data are complemented by simulations using a simple spring-block model where the asperity contact regions are modeled by miniblocks connected to the big block by viscoelastic springs. The miniblocks experience forces that fluctuate randomly with the lateral position, simulating the interaction between asperities on the block and the substrate. The theoretical model provides displacement power spectra that agree well with the experimental results.
{"title":"Brownian friction dynamics: fluctuations in sliding distance","authors":"Ruibin Xu, Feng Zhou, B. N. J. Persson","doi":"arxiv-2406.16139","DOIUrl":"https://doi.org/arxiv-2406.16139","url":null,"abstract":"We have studied the fluctuation (noise) in the position of sliding blocks\u0000under constant driving forces on different substrate surfaces. The experimental\u0000data are complemented by simulations using a simple spring-block model where\u0000the asperity contact regions are modeled by miniblocks connected to the big\u0000block by viscoelastic springs. The miniblocks experience forces that fluctuate\u0000randomly with the lateral position, simulating the interaction between\u0000asperities on the block and the substrate. The theoretical model provides\u0000displacement power spectra that agree well with the experimental results.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"76 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141500974","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Stochastic dynamics on sparse graphs and disordered systems often lead to complex behaviors characterized by heterogeneity in time and spatial scales, slow relaxation, localization, and aging phenomena. The mathematical tools and approximation techniques required to analyze these complex systems are still under development, posing significant technical challenges and resulting in a reliance on numerical simulations. We introduce a novel computational framework for investigating the dynamics of sparse disordered systems with continuous degrees of freedom. Starting with a graphical model representation of the dynamic partition function for a system of linearly-coupled stochastic differential equations, we use dynamic cavity equations on locally tree-like factor graphs to approximate the stochastic measure. Here, cavity marginals are identified with local functionals of single-site trajectories. Our primary approximation involves a second-order truncation of a small-coupling expansion, leading to a Gaussian form for the cavity marginals. For linear dynamics with additive noise, this method yields a closed set of causal integro-differential equations for cavity versions of one-time and two-time averages. These equations provide an exact dynamical description within the local tree-like approximation, retrieving classical results for the spectral density of sparse random matrices. Global constraints, non-linear forces, and state-dependent noise terms can be addressed using a self-consistent perturbative closure technique. The resulting equations resemble those of dynamical mean-field theory in the mode-coupling approximation used for fully-connected models. However, due to their cavity formulation, the present method can also be applied to ensembles of sparse random graphs and employed as a message-passing algorithm on specific graph instances.
{"title":"Gaussian approximation of dynamic cavity equations for linearly-coupled stochastic dynamics","authors":"Mattia Tarabolo, Luca Dall'Asta","doi":"arxiv-2406.14200","DOIUrl":"https://doi.org/arxiv-2406.14200","url":null,"abstract":"Stochastic dynamics on sparse graphs and disordered systems often lead to\u0000complex behaviors characterized by heterogeneity in time and spatial scales,\u0000slow relaxation, localization, and aging phenomena. The mathematical tools and\u0000approximation techniques required to analyze these complex systems are still\u0000under development, posing significant technical challenges and resulting in a\u0000reliance on numerical simulations. We introduce a novel computational framework\u0000for investigating the dynamics of sparse disordered systems with continuous\u0000degrees of freedom. Starting with a graphical model representation of the\u0000dynamic partition function for a system of linearly-coupled stochastic\u0000differential equations, we use dynamic cavity equations on locally tree-like\u0000factor graphs to approximate the stochastic measure. Here, cavity marginals are\u0000identified with local functionals of single-site trajectories. Our primary\u0000approximation involves a second-order truncation of a small-coupling expansion,\u0000leading to a Gaussian form for the cavity marginals. For linear dynamics with\u0000additive noise, this method yields a closed set of causal integro-differential\u0000equations for cavity versions of one-time and two-time averages. These\u0000equations provide an exact dynamical description within the local tree-like\u0000approximation, retrieving classical results for the spectral density of sparse\u0000random matrices. Global constraints, non-linear forces, and state-dependent\u0000noise terms can be addressed using a self-consistent perturbative closure\u0000technique. The resulting equations resemble those of dynamical mean-field\u0000theory in the mode-coupling approximation used for fully-connected models.\u0000However, due to their cavity formulation, the present method can also be\u0000applied to ensembles of sparse random graphs and employed as a message-passing\u0000algorithm on specific graph instances.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"65 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141500976","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Perovskite Quantum Dots (PQDs) have a promising future for several applications due to their unique properties. This study investigates the effectiveness of Machine Learning (ML) in predicting the size, absorbance (1S abs) and photoluminescence (PL) properties of $mathrm{CsPbCl}_3$ PQDs using synthesizing features as the input dataset. the study employed ML models of Support Vector Regression (SVR), Nearest Neighbour Distance (NND), Random Forest (RF), Gradient Boosting Machine (GBM), Decision Tree (DT) and Deep Learning (DL). Although all models performed highly accurate results, SVR and NND demonstrated the best accurate property prediction by achieving excellent performance on the test and training datasets, with high $mathrm{R}^2$ and low Root Mean Squared Error (RMSE) and low Mean Absolute Error (MAE) metric values. Given that ML is becoming more superior, its ability to understand the QDs field could prove invaluable to shape the future of nanomaterials designing.
包光体量子点（PQDs）因其独特的性质，在多种应用中具有广阔的前景。本研究以合成特征作为输入数据集，探讨了机器学习（ML）在预测$mathrm{CsPbCl}_3$ PQDs的尺寸、吸光度（1Sabs）和光致发光（PL）特性方面的有效性。该研究采用了支持向量回归（SVR）、近邻距离（NND）、随机森林（RF）、梯度提升机（GBM）、决策树（DT）和深度学习（DL）等 ML 模型。尽管所有模型都取得了非常准确的结果，但SVR和NND在测试和训练数据集上取得了优异的性能，具有较高的$mathrm{R}^2$、较低的根均方误差（RMSE）和较低的平均绝对误差（MAE）指标值，从而展示了最准确的性能预测。
{"title":"Machine Learning Models for Accurately Predicting Properties of CsPbCl3 Perovskite Quantum Dots","authors":"Mehmet Sıddık Çadırcı, Musa Çadırcı","doi":"arxiv-2406.15515","DOIUrl":"https://doi.org/arxiv-2406.15515","url":null,"abstract":"Perovskite Quantum Dots (PQDs) have a promising future for several\u0000applications due to their unique properties. This study investigates the\u0000effectiveness of Machine Learning (ML) in predicting the size, absorbance (1S\u0000abs) and photoluminescence (PL) properties of $mathrm{CsPbCl}_3$ PQDs using\u0000synthesizing features as the input dataset. the study employed ML models of\u0000Support Vector Regression (SVR), Nearest Neighbour Distance (NND), Random\u0000Forest (RF), Gradient Boosting Machine (GBM), Decision Tree (DT) and Deep\u0000Learning (DL). Although all models performed highly accurate results, SVR and\u0000NND demonstrated the best accurate property prediction by achieving excellent\u0000performance on the test and training datasets, with high $mathrm{R}^2$ and low\u0000Root Mean Squared Error (RMSE) and low Mean Absolute Error (MAE) metric values.\u0000Given that ML is becoming more superior, its ability to understand the QDs\u0000field could prove invaluable to shape the future of nanomaterials designing.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"11 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141500973","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jalmari Passilahti, Anton Vladyka, Johannes Niskanen
Encoder-decoder neural networks (EDNN) condense information most relevant to the output of the feedforward network to activation values at a bottleneck layer. We study the use of this architecture in emulation and interpretation of simulated X-ray spectroscopic data with the aim to identify key structural characteristics for the spectra, previously studied using emulator-based component analysis (ECA). We find an EDNN to outperform ECA in covered target variable variance, but also discover complications in interpreting the latent variables in physical terms. As a compromise of the benefits of these two approaches, we develop a network where the linear projection of ECA is used, thus maintaining the beneficial characteristics of vector expansion from the latent variables for their interpretation. These results underline the necessity of information recovery after its condensation and identification of decisive structural degrees for the output spectra for a justified interpretation.
编码器-解码器神经网络（EDNN）将与前馈网络输出最相关的信息浓缩为瓶颈层的激活值。我们研究了这种结构在模拟和解释模拟 X 射线光谱数据中的应用，目的是识别光谱的关键结构特征。我们发现 EDNN 在覆盖目标变量方差方面优于 ECA，但也发现了用物理术语解释潜变量的复杂性。为了折中这两种方法的优点，我们开发了一种使用 ECA 线性投影的网络，从而保持了从潜在变量向量扩展来解释潜在变量的有利特性。这些结果凸显了在信息浓缩后进行信息恢复的必要性，以及为输出光谱确定决定性结构度以进行合理解释的必要性。
{"title":"Encoder-Decoder Neural Networks in Interpretation of X-ray Spectra","authors":"Jalmari Passilahti, Anton Vladyka, Johannes Niskanen","doi":"arxiv-2406.14044","DOIUrl":"https://doi.org/arxiv-2406.14044","url":null,"abstract":"Encoder-decoder neural networks (EDNN) condense information most relevant to\u0000the output of the feedforward network to activation values at a bottleneck\u0000layer. We study the use of this architecture in emulation and interpretation of\u0000simulated X-ray spectroscopic data with the aim to identify key structural\u0000characteristics for the spectra, previously studied using emulator-based\u0000component analysis (ECA). We find an EDNN to outperform ECA in covered target\u0000variable variance, but also discover complications in interpreting the latent\u0000variables in physical terms. As a compromise of the benefits of these two\u0000approaches, we develop a network where the linear projection of ECA is used,\u0000thus maintaining the beneficial characteristics of vector expansion from the\u0000latent variables for their interpretation. These results underline the\u0000necessity of information recovery after its condensation and identification of\u0000decisive structural degrees for the output spectra for a justified\u0000interpretation.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"20 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141500978","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We consider the problem of a graph subjected to adversarial perturbations, such as those arising from cyber-attacks, where edges are covertly added or removed. The adversarial perturbations occur during the transmission of the graph between a sender and a receiver. To counteract potential perturbations, we explore a repetition coding scheme with sender-assigned binary noise and majority voting on the receiver's end to rectify the graph's structure. Our approach operates without prior knowledge of the attack's characteristics. We provide an analytical derivation of a bound on the number of repetitions needed to satisfy probabilistic constraints on the quality of the reconstructed graph. We show that the method can accurately decode graphs that were subjected to non-random edge removal, namely, those connected to vertices with the highest eigenvector centrality, in addition to random addition and removal of edges by the attacker.
{"title":"On countering adversarial perturbations in graphs using error correcting codes","authors":"Saif Eddin Jabari","doi":"arxiv-2406.14245","DOIUrl":"https://doi.org/arxiv-2406.14245","url":null,"abstract":"We consider the problem of a graph subjected to adversarial perturbations,\u0000such as those arising from cyber-attacks, where edges are covertly added or\u0000removed. The adversarial perturbations occur during the transmission of the\u0000graph between a sender and a receiver. To counteract potential perturbations,\u0000we explore a repetition coding scheme with sender-assigned binary noise and\u0000majority voting on the receiver's end to rectify the graph's structure. Our\u0000approach operates without prior knowledge of the attack's characteristics. We\u0000provide an analytical derivation of a bound on the number of repetitions needed\u0000to satisfy probabilistic constraints on the quality of the reconstructed graph.\u0000We show that the method can accurately decode graphs that were subjected to\u0000non-random edge removal, namely, those connected to vertices with the highest\u0000eigenvector centrality, in addition to random addition and removal of edges by\u0000the attacker.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"27 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141500975","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Anastasia Ragulskaya, Vladimir Starostin, Fajun Zhang, Christian Gutt, Frank Schreiber
X-ray photon correlation spectroscopy (XPCS) is a powerful tool for the investigation of dynamics covering a broad range of time and length scales. The two-time correlation function (TTC) is commonly used to track non-equilibrium dynamical evolution in XPCS measurements, followed by the extraction of one-time correlations. While the theoretical foundation for the quantitative analysis of TTCs is primarily established for equilibrium systems, where key parameters such as diffusion remain constant, non-equilibrium systems pose a unique challenge. In such systems, different projections ("cuts") of the TTC may lead to divergent results if the underlying fundamental parameters themselves are subject to temporal variations. This article explores widely used approaches for TTC calculations and common methods for extracting relevant information from correlation functions on case studies, particularly in the light of comparing dynamics in equilibrium and non-equilibrium systems.
X 射线光子相关光谱学（XPCS）是研究涵盖广泛时间和长度尺度的动力学的有力工具。在 XPCS 测量中，双时间相关函数（TTC）通常用于跟踪非平衡动力学演化，然后提取单时间相关性。虽然定量分析 TTC 的理论基础主要是针对平衡系统建立的，在平衡系统中，扩散等关键参数保持不变，但非平衡系统却带来了独特的挑战。在这类系统中，如果基本参数本身会发生时间变化，那么对 TTC 的不同预测（"切割"）可能会导致不同的结果。本文探讨了广泛使用的 TTC 计算方法，以及从案例研究的相关函数中提取相关信息的常用方法，特别是在比较平衡和非平衡系统的动力学方面。
{"title":"On the analysis of two-time correlation functions: equilibrium vs non-equilibrium systems","authors":"Anastasia Ragulskaya, Vladimir Starostin, Fajun Zhang, Christian Gutt, Frank Schreiber","doi":"arxiv-2406.12520","DOIUrl":"https://doi.org/arxiv-2406.12520","url":null,"abstract":"X-ray photon correlation spectroscopy (XPCS) is a powerful tool for the\u0000investigation of dynamics covering a broad range of time and length scales. The\u0000two-time correlation function (TTC) is commonly used to track non-equilibrium\u0000dynamical evolution in XPCS measurements, followed by the extraction of\u0000one-time correlations. While the theoretical foundation for the quantitative\u0000analysis of TTCs is primarily established for equilibrium systems, where key\u0000parameters such as diffusion remain constant, non-equilibrium systems pose a\u0000unique challenge. In such systems, different projections (\"cuts\") of the TTC\u0000may lead to divergent results if the underlying fundamental parameters\u0000themselves are subject to temporal variations. This article explores widely\u0000used approaches for TTC calculations and common methods for extracting relevant\u0000information from correlation functions on case studies, particularly in the\u0000light of comparing dynamics in equilibrium and non-equilibrium systems.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"35 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141500977","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The ROC (receiver operating characteristic) curve is a widely used device for assessing decision-making systems. It seems surprising, in view of its history dating back to World War Two, that the assignment of uncertainties to a ROC curve is apparently not settled. This note returns to the question, focusing on the application of ROC curves to the analysis of data from counting experiments and taking a practical operational approach to the concept of uncertainty.
{"title":"Uncertainties in ROC (Receiver Operating Characteristic) Curves Derived from Counting Data","authors":"M. P. Fewell","doi":"arxiv-2406.11396","DOIUrl":"https://doi.org/arxiv-2406.11396","url":null,"abstract":"The ROC (receiver operating characteristic) curve is a widely used device for\u0000assessing decision-making systems. It seems surprising, in view of its history\u0000dating back to World War Two, that the assignment of uncertainties to a ROC\u0000curve is apparently not settled. This note returns to the question, focusing on\u0000the application of ROC curves to the analysis of data from counting experiments\u0000and taking a practical operational approach to the concept of uncertainty.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"32 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141500979","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The advancement of technology has led to rampant growth in data collection across almost every field, including astrophysics, with researchers turning to machine learning to process and analyze this data. One prominent example of this data in astrophysics is the atmospheric retrievals of exoplanets. In order to help bridge the gap between machine learning and astrophysics domain experts, the 2023 Ariel Data Challenge was hosted to predict posterior distributions of 7 exoplanetary features. The procedure outlined in this paper leveraged a combination of two deep learning models to address this challenge: a Multivariate Gaussian model that generates the mean and covariance matrix of a multivariate Gaussian distribution, and a Uniform Quantile model that predicts quantiles for use as the upper and lower bounds of a uniform distribution. Training of the Multivariate Gaussian model was found to be unstable, while training of the Uniform Quantile model was stable. An ensemble of uniform distributions was found to have competitive results during testing (posterior score of 696.43), and when combined with a multivariate Gaussian distribution achieved a final rank of third in the 2023 Ariel Data Challenge (final score of 681.57).
随着技术的进步，几乎所有领域（包括天体物理学）的数据收集量都在急剧增长，研究人员转而利用机器学习来处理和分析这些数据。天体物理学中的一个突出例子就是系外行星的大气检索数据。为了帮助缩小机器学习与天体物理学领域专家之间的差距，2023 年阿里尔数据挑战赛（Ariel Data Challenge）旨在预测 7 个系外行星特征的后验分布。本文概述的程序利用了两个深度学习模型的组合来应对这一挑战：一个是多变量高斯模型，用于生成多变量高斯分布的均值和协方差矩阵；另一个是均匀量值模型，用于预测作为均匀分布上下限的量值。多变量高斯模型的训练并不稳定，而均匀量值模型的训练则很稳定。在测试过程中，发现均匀分布集合具有竞争力的结果（后验得分为 696.43），当与多元高斯分布结合时，在 2023 年阿里尔数据挑战赛中取得了第三名的最终排名（最终得分为 681.57）。
{"title":"Predicting Exoplanetary Features with a Residual Model for Uniform and Gaussian Distributions","authors":"Andrew Sweet","doi":"arxiv-2406.10771","DOIUrl":"https://doi.org/arxiv-2406.10771","url":null,"abstract":"The advancement of technology has led to rampant growth in data collection\u0000across almost every field, including astrophysics, with researchers turning to\u0000machine learning to process and analyze this data. One prominent example of\u0000this data in astrophysics is the atmospheric retrievals of exoplanets. In order\u0000to help bridge the gap between machine learning and astrophysics domain\u0000experts, the 2023 Ariel Data Challenge was hosted to predict posterior\u0000distributions of 7 exoplanetary features. The procedure outlined in this paper\u0000leveraged a combination of two deep learning models to address this challenge:\u0000a Multivariate Gaussian model that generates the mean and covariance matrix of\u0000a multivariate Gaussian distribution, and a Uniform Quantile model that\u0000predicts quantiles for use as the upper and lower bounds of a uniform\u0000distribution. Training of the Multivariate Gaussian model was found to be\u0000unstable, while training of the Uniform Quantile model was stable. An ensemble\u0000of uniform distributions was found to have competitive results during testing\u0000(posterior score of 696.43), and when combined with a multivariate Gaussian\u0000distribution achieved a final rank of third in the 2023 Ariel Data Challenge\u0000(final score of 681.57).","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"19 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141500980","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Daniel E. Lopez-Fogliani, Andres D. Perez, Roberto Ruiz de Austri
The detection of Dark Matter (DM) remains a significant challenge in particle physics. This study exploits advanced machine learning models to improve detection capabilities of liquid xenon time projection chamber experiments, utilizing state-of-the-art transformers alongside traditional methods like Multilayer Perceptrons and Convolutional Neural Networks. We evaluate various data representations and find that simplified feature representations, particularly corrected S1 and S2 signals, retain critical information for classification. Our results show that while transformers offer promising performance, simpler models like XGBoost can achieve comparable results with optimal data representations. We also derive exclusion limits in the cross-section versus DM mass parameter space, showing minimal differences between XGBoost and the best performing deep learning models. The comparative analysis of different machine learning approaches provides a valuable reference for future experiments by guiding the choice of models and data representations to maximize detection capabilities.
{"title":"Insights into Dark Matter Direct Detection Experiments: Decision Trees versus Deep Learning","authors":"Daniel E. Lopez-Fogliani, Andres D. Perez, Roberto Ruiz de Austri","doi":"arxiv-2406.10372","DOIUrl":"https://doi.org/arxiv-2406.10372","url":null,"abstract":"The detection of Dark Matter (DM) remains a significant challenge in particle\u0000physics. This study exploits advanced machine learning models to improve\u0000detection capabilities of liquid xenon time projection chamber experiments,\u0000utilizing state-of-the-art transformers alongside traditional methods like\u0000Multilayer Perceptrons and Convolutional Neural Networks. We evaluate various\u0000data representations and find that simplified feature representations,\u0000particularly corrected S1 and S2 signals, retain critical information for\u0000classification. Our results show that while transformers offer promising\u0000performance, simpler models like XGBoost can achieve comparable results with\u0000optimal data representations. We also derive exclusion limits in the\u0000cross-section versus DM mass parameter space, showing minimal differences\u0000between XGBoost and the best performing deep learning models. The comparative\u0000analysis of different machine learning approaches provides a valuable reference\u0000for future experiments by guiding the choice of models and data representations\u0000to maximize detection capabilities.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"73 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141500981","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：