ANNZ plus : an enhanced photometric redshift estimation algorithm with applications on the PAU survey

被引：0

作者：

Pathi, Imdad Mahmud ^{[1
]}

Soo, John Y. H. ^{[1
]}

Wee, Mao Jie ^{[1
]}

Zakaria, Sazatul Nadhilah ^{[1
]}

Ismail, Nur Azwin ^{[1
]}

Baugh, Carlton M. ^{[2
]}

Manzoni, Giorgio ^{[3
]}

Gaztanaga, Enrique ^{[4
,5
,6
]}

Castander, Francisco J. ^{[5
,6
]}

Eriksen, Martin ^{[7
,8
]}

Carretero, Jorge ^{[8
,9
]}

Fernandez, Enrique ^{[7
]}

Garcia-Bellido, Juan ^{[10
]}

Miquel, Ramon ^{[7
,11
]}

Padilla, Cristobal ^{[7
]}

Renard, Pablo ^{[12
]}

Sanchez, Eusebio ^{[9
]}

Sevilla-Noarbe, Ignacio ^{[9
]}

Tallada-Crespi, Pau ^{[8
,9
]}

机构：

[1] Univ Sains Malaysia, Sch Phys, Usm 11800, Pulau Pinang, Malaysia

[2] Univ Durham, Inst Computat Cosmol, Dept Phys, Sci Labs, South Rd, Durham DH1 3LE, England

[3] Hong Kong Univ Sci & Technol, Jockey Club Inst Adv Study, Hong Kong, Peoples R China

[4] Univ Portsmouth, Inst Cosmol & Gravitat ICG, Portsmouth PO1 3FX, England

[5] Univ Autonoma Barcelona, Inst Space Sci, CSIC, ICE, Carrer de Can Magrans S-N, E-08193 Cerdanyola Del Valles, Barcelona, Spain

[6] Univ Politecn Cataluna, Inst Estudis Espacials Catalunya IEEC, Edifci RDIT, E-08860 Castelldefels, Barcelona, Spain

[7] Univ Autonoma Barcelona, Inst Fis Altes Energies IFAE, E-08193 Bellaterra, Barcelona, Spain

[8] Univ Autonoma Barcelona, Port Informacio Cient PIC, Carrer Albareda S-N, E-08193 Bellaterra, Barcelona, Spain

[9] Ctr Invest Energet Medioambientales & Tecnol CIEMA, Ave Complutense 40, E-28040 Madrid, Spain

[10] Univ Autonoma Madrid, Inst Fis Teor, CSIC, E-28049 Canto Blanco, Madrid, Spain

[11] Inst Catalana Rercerca & Estudis Avancats ICREA, Pg de Lluis Co 23, E-08010 Barcelona, Spain

[12] Tsinghua Univ, Dept Astron, Beijing 100084, Peoples R China

来源：

JOURNAL OF COSMOLOGY AND ASTROPARTICLE PHYSICS | 2025年 / 01期

基金：

美国安德鲁·梅隆基金会;

关键词：

galaxy surveys; high redshift galaxies; Machine learning; DIGITAL SKY SURVEY; DEEP NEURAL-NETWORKS; GALAXY FORMATION; MACHINE; SELECTION; CATALOG; MODELS;

D O I：

10.1088/1475-7516/2025/01/097

中图分类号：

P1 [天文学];

学科分类号：

0704 ;

摘要：

ANNz is a fast and simple algorithm which utilises artificial neural networks (ANNs), it was known as one of the pioneers of machine learning approaches to photometric redshift estimation decades ago. We enhanced the algorithm by introducing new activation functions like tanh, softplus, SiLU, Mish and ReLU variants; its new performance is then vigorously tested on legacy samples like the Luminous Red Galaxy (LRG) and Stripe-82 samples from SDSS, as well as modern galaxy samples like the Physics of the Accelerating Universe Survey (PAUS). This work focuses on testing the robustness of activation functions with respect to the choice of ANN architectures, particularly on its depth and width, in the context of galaxy photometric redshift estimation. Our upgraded algorithm, which we named ANNz+, shows that the tanh and Leaky ReLU activation functions provide more consistent and stable results across deeper and wider architectures with > 1 per cent improvement in root-mean-square error ( A RMS ) and 68th percentile error (A68) when tested on SDSS data sets. While assessing its capabilities in handling high dimensional inputs, we achieved an improvement of 11 per cent in A RMS and 6 per cent in A 68 with the tanh activation function when tested on the 40-narrowband PAUS dataset; it even outperformed ANNz2, its supposed successor, by 44 per cent in A RMS . This justifies the effort to upgrade the 20-year-old ANNz, allowing it to remain viable and competitive within the photo-z community today. The updated algorithm ANNz+ is publicly available at https://github.com/imdadmpt/ANNzPlus.

引用

页数：41