Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Improving the discovery of near-Earth objects with machine-learning methods

Peter Vereš, Richard Cloete, Matthew J. Payne, Abraham Loeb

TL;DR

This work tackles the persistent challenge of high false positives on the NEO Confirmation Page (NEOCP) by analyzing the digest2 parameter space and applying four machine-learning classifiers to distinguish NEOs from non-NEOs using 2019–2024 data. A set of digest2-based filters ($filter_{0}$, $filter_{1}$, $filter_{2}$) is developed and evaluated, achieving substantial non-NEO reduction while limiting NEO losses; in tandem, SGD-based classification (and ensembles) yields an overall accuracy around $0.93$ and high NEO precision around $0.95$–$0.98$, enabling an effective net reduction in follow-up burden. The combined approach reduces non-NEO postings by over $80 ext{%}$ with only about $5.5 ext{%}$ loss of NEO tracklets, and most initially misclassified NEOs are later recovered through follow-up, suggesting a practical path to reallocate observing resources toward high-priority objects. The study also demonstrates that ML models can adapt to changes in surveys and data volumes, with data and tools openly available to support reproducibility and future retraining.”

Abstract

We present a comprehensive analysis of the digest2 parameters for candidates of the Near-Earth Object Confirmation Page (NEOCP) that were reported between 2019 and 2024. Our study proposes methods for significantly reducing the inclusion of non-NEO objects on the NEOCP. Despite the substantial increase in near-Earth object (NEO) discoveries in recent years, only about half of the NEOCP candidates are ultimately confirmed as NEOs. Therefore, much observing time is spent following up on non-NEOs. Furthermore, approximately 11% of the candidates remain unconfirmed because the follow-up observations are insufficient. These are nearly 600 cases per year. To reduce false positives and minimize wasted resources on non-NEOs, we refine the posting criteria for NEOCP based on a detailed analysis of all digest2 scores. We investigated 30 distinct digest2 parameter categories for candidates that were confirmed as NEOs and non-NEOs. From this analysis, we derived a filtering mechanism based on selected digest2 parameters that were able to exclude 20% of the non-NEOs from the NEOCP while maintaining a minimal loss of true NEOs. We also investigated the application of four machine-learning (ML) techniques, that is, the gradient-boosting machine (GBM), the random forest (RF) classifier, the stochastic gradient descent (SGD) classifier, and neural networks (NN) to classify NEOCP candidates as NEOs or non-NEOs. Based on digest2 parameters as input, our ML models achieved a precision of approximately 95% in distinguishing between NEOs and non-NEOs. Results. Combining the digest2 parameter filter with an ML-based classification model, we demonstrate a significant reduction in non-NEOs on the NEOCP that exceeds 80%, while limiting the loss of NEO discovery tracklets to 5.5%. Importantly, we show that most follow-up tracklets of initially misclassified NEOs are later correctly identified as NEOs.

Improving the discovery of near-Earth objects with machine-learning methods

TL;DR

This work tackles the persistent challenge of high false positives on the NEO Confirmation Page (NEOCP) by analyzing the digest2 parameter space and applying four machine-learning classifiers to distinguish NEOs from non-NEOs using 2019–2024 data. A set of digest2-based filters (, , ) is developed and evaluated, achieving substantial non-NEO reduction while limiting NEO losses; in tandem, SGD-based classification (and ensembles) yields an overall accuracy around and high NEO precision around –, enabling an effective net reduction in follow-up burden. The combined approach reduces non-NEO postings by over with only about loss of NEO tracklets, and most initially misclassified NEOs are later recovered through follow-up, suggesting a practical path to reallocate observing resources toward high-priority objects. The study also demonstrates that ML models can adapt to changes in surveys and data volumes, with data and tools openly available to support reproducibility and future retraining.”

Abstract

We present a comprehensive analysis of the digest2 parameters for candidates of the Near-Earth Object Confirmation Page (NEOCP) that were reported between 2019 and 2024. Our study proposes methods for significantly reducing the inclusion of non-NEO objects on the NEOCP. Despite the substantial increase in near-Earth object (NEO) discoveries in recent years, only about half of the NEOCP candidates are ultimately confirmed as NEOs. Therefore, much observing time is spent following up on non-NEOs. Furthermore, approximately 11% of the candidates remain unconfirmed because the follow-up observations are insufficient. These are nearly 600 cases per year. To reduce false positives and minimize wasted resources on non-NEOs, we refine the posting criteria for NEOCP based on a detailed analysis of all digest2 scores. We investigated 30 distinct digest2 parameter categories for candidates that were confirmed as NEOs and non-NEOs. From this analysis, we derived a filtering mechanism based on selected digest2 parameters that were able to exclude 20% of the non-NEOs from the NEOCP while maintaining a minimal loss of true NEOs. We also investigated the application of four machine-learning (ML) techniques, that is, the gradient-boosting machine (GBM), the random forest (RF) classifier, the stochastic gradient descent (SGD) classifier, and neural networks (NN) to classify NEOCP candidates as NEOs or non-NEOs. Based on digest2 parameters as input, our ML models achieved a precision of approximately 95% in distinguishing between NEOs and non-NEOs. Results. Combining the digest2 parameter filter with an ML-based classification model, we demonstrate a significant reduction in non-NEOs on the NEOCP that exceeds 80%, while limiting the loss of NEO discovery tracklets to 5.5%. Importantly, we show that most follow-up tracklets of initially misclassified NEOs are later correctly identified as NEOs.
Paper Structure (13 sections, 17 figures, 11 tables)

This paper contains 13 sections, 17 figures, 11 tables.

Table of Contents

  1. Introduction
  2. NEOCP data
  3. Digest2
  4. Digest2 analysis of NEO candidates
  5. Machine-learning methods
  6. Data preprocessing
  7. Model architecture and implementation
  8. Results
  9. Decreasing the false-positive rate
  10. Follow-up tracklets of misidentified NEOs by filtering.
  11. Analysis of unconfirmed candidates
  12. Discussion
  13. Data availability

Figures (17)

  • Figure 1: Number of NEO discoveries per year. Large NEOs with a size of > 1 km (H<18) and the NASA target (H<22) are displayed as well.
  • Figure 2: Final disposition of the NEOCP tracklets between March 14, 2019 and December 31, 2024.
  • Figure 3: Number of NEOCP candidates posted per year.
  • Figure 4: Most productive observatory codes on the NEOCP between 2019-2024. The ATLAS survey has four telescopes: T05, T08, W68, and M22. Pan-STARRS (PS) has two telescopes: F51 and F52. The Catalina Sky Survey (CSS) has five telescopes: G96, 703, I52, V00, and V06.
  • Figure 5: Fraction of NEOs on NEOCP in the NEO noid digest2 score (left) and the score of NEOs and non-NEOs on NEOCP in 2019-2024.
  • ...and 12 more figures