Probabilistic Random Forest: A machine learning algorithm for noisy datasets

Abstract

Machine learning (ML) algorithms become increasingly important in the analysis of astronomical data. However, since most ML algorithms are not designed to take data uncertainties into account, ML based studies are mostly restricted to data with high signal-to-noise ratio. Astronomical datasets of such high-quality are uncommon. In this work we modify the long-established Random Forest (RF) algorithm to take into account uncertainties in the measurements (i.e., features) as well as in the assigned classes (i.e., labels). To do so, the Probabilistic Random Forest (PRF) algorithm treats the features and labels as probability distribution functions, rather than deterministic quantities. We perform a variety of experiments where we inject different types of noise to a dataset, and compare the accuracy of the PRF to that of RF. The PRF outperforms RF in all cases, with a moderate increase in running time. We find an improvement in classification accuracy of up to 10% in the case of noisy features, and up to 30% in the case of noisy labels. The PRF accuracy decreased by less then 5% for a dataset with as many as 45% misclassified objects, compared to a clean dataset. Apart from improving the prediction accuracy in noisy datasets, the PRF naturally copes with missing values in the data, and outperforms RF when applied to a dataset with different noise characteristics in the training and test sets, suggesting that it can be used for Transfer Learning.

Accepted by AJ, comments are welcome! Code is available at https://github.com/ireis/PRF

References in corpus (33)

• Scikit-learn: Machine Learning in Python
• Galaxy Zoo : Morphologies derived from visual inspection of galaxies from the Sloan Digital Sky Survey
• Unification of Luminous Type 1 Quasars through CIV Emission
• Generative Adversarial Networks recover features in astrophysical images of galaxies beyond the deconvolution limit
• Galaxy Zoo: Reproducing Galaxy Morphologies Via Machine Learning
• Photometric redshift estimation via deep learning
• Automating Discovery and Classification of Transients and Variable Stars in the Synoptic Survey Era
• The weirdest SDSS galaxies: results from an outlier detection algorithm
• A recurrent neural network for classification of unevenly sampled variable stars
• Using Machine Learning for Discovery in Synoptic Survey Imaging
• Deep Learning of Quasar Spectra to Discover and Characterize Damped Lya Systems
• Finding outlier light-curves in catalogs of periodic variable stars
• Photometric classification of type Ia supernovae in the SuperNova Legacy Survey with supervised learning
• Shallow Transits - Deep Learning I: Feasibility Study of Deep Learning to Detect Periodic Transits of Exoplanets
• MADE: A spectroscopic Mass, Age, and Distance Estimator for red giant stars with Bayesian machine learning
• Unusual quasars from the Sloan Digital Sky Survey selected by means of Kohonen self-organising maps
• A Hybrid Ensemble Learning Approach to Star-Galaxy Classification
• Automated Clustering Algorithms for Classification of Astronomical Objects
• Automated Classification of Periodic Variable Stars detected by the Wide-field Infrared Survey Explorer
• Detecting outliers and learning complex structures with large spectroscopic surveys - a case study with APOGEE stars
• Automatic Classification of Variable Stars in Catalogs with missing data
• Return of the features. Efficient feature selection and interpretation for photometric redshifts
• An improved quasar detection method in EROS-2 and MACHO LMC datasets
• Using machine learning to classify the diffuse interstellar bands
• Preparing for advanced LIGO: A Star-Galaxy Separation Catalog for the Palomar Transient Factory
• Uncertain classification of Variable Stars: handling observational GAPS and noise
• Minimax fast rates for discriminant analysis with errors in variables
• Using the Properties of Broad Absorption Line Quasars to Illuminate Quasar Structure
• Random Forest Classification of Stars in the Galactic Centre
• Redshifted broad absorption line quasars found via machine-learned spectral similarity
• Automated Real-Time Classification and Decision Making in Massive Data Streams from Synoptic Sky Surveys
• An Explorative Approach for Inspecting Kepler Data
• Model-Coupled Autoencoder for Time Series Visualisation

Cited by in corpus (33)

• Surveying the reach and maturity of machine learning and artificial intelligence in astronomy
• Optical Fiber Fault Detection and Localization in a Noisy OTDR Trace Based on Denoising Convolutional Autoencoder and Bidirectional Long Short-Term Memory
• Classifying Exoplanet Candidates with Convolutional Neural Networks: Application to the Next Generation Transit Survey
• The Gaia EDR3 view of Johnson-Kron-Cousins standard stars: the curated Landolt and Stetson collections
• Blind chemical tagging with DBSCAN: prospects for spectroscopic surveys
• Photometric redshifts with machine learning, lights and shadows on a complex data science use case
• Deep Neural Network Classifier for Variable Stars with Novelty Detection Capability
• Photometric redshifts for X-ray-selected active galactic nuclei in the eROSITA era
• A Survey for High-redshift Gravitationally Lensed Quasars and Close Quasars Pairs. I. the Discoveries of an Intermediately-lensed Quasar and a Kpc-scale Quasar Pair at $z\sim 5$
• Spectroscopy of QUBRICS quasar candidates: 1672 new redshifts and a Golden Sample for the Sandage Test of the Redshift Drift
• Stellar population of the Rosette Nebula and NGC 2244: application of the probabilistic random forest
• Computational Tools for the Spectroscopic Analysis of White Dwarfs
• The Probabilistic Random Forest applied to the selection of quasar candidates in the QUBRICS Survey
• UOCS. III. UVIT catalogue of open clusters with machine learning based membership using \textit{Gaia} EDR3 astrometry
• Searching for AGN and Pulsar Candidates in 4FGL Unassociated Sources Using Machine Learning
• Massive young stellar objects in the Local Group irregular galaxy NGC6822 identified using machine learning
• A new automated tool for the spectral classification of OB stars
• The miniJPAS survey quasar selection II: Machine learning classification with photometric measurements and uncertainties
• Massive young stellar objects in the Local Group spiral galaxy M33 identified using machine learning
• Selection of powerful radio galaxies with machine learning
• Applied Machine-Learning Models to Identify Spectral Sub-Types of M Dwarfs from Photometric Surveys
• Artificial Neural Networks for cosmic gamma-ray propagation in the Universe
• Anomaly detection in Astrophysics: a comparison between unsupervised Deep and Machine Learning on KiDS data
• The probabilistic random forest applied to the QUBRICS survey: improving the selection of high-redshift quasars with synthetic data
• Real-time Light Curve Classification Framework for the Wide Field Survey Telescope Using Modified Semi-supervised Variational Auto-Encoder
• Machine Learning Classification of Young Stellar Objects and Evolved Stars in the Magellanic Clouds Using the Probabilistic Random Forest Classifier
• yonder: A python package for data denoising and reconstruction
• Galactic Component Mapping of Galaxy UGC 2885 by Machine Learning Classification
• Success of Machine Learning algorithms in Dynamical Mass Measurements of Galaxy Clusters
• Discovering Exotic AGN behind the Magellanic Clouds
• A machine learning approach to estimating HI deficiency in galaxies
• Estimating the completeness of the QUBRICS Survey with 3501 QSO redshifts from Gaia DR3 spectra
• Internal dynamics and structure of Cepheus OB4. The asymmetric expansion of Berkeley 59