-
Supervised Radio Frequency Interference Detection with SNNs
Authors:
Nicholas J. Pritchard,
Andreas Wicenec,
Mohammed Bennamoun,
Richard Dodson
Abstract:
Radio Frequency Interference (RFI) poses a significant challenge in radio astronomy, arising from terrestrial and celestial sources, disrupting observations conducted by radio telescopes. Addressing RFI involves intricate heuristic algorithms, manual examination, and, increasingly, machine learning methods. Given the dynamic and temporal nature of radio astronomy observations, Spiking Neural Netwo…
▽ More
Radio Frequency Interference (RFI) poses a significant challenge in radio astronomy, arising from terrestrial and celestial sources, disrupting observations conducted by radio telescopes. Addressing RFI involves intricate heuristic algorithms, manual examination, and, increasingly, machine learning methods. Given the dynamic and temporal nature of radio astronomy observations, Spiking Neural Networks (SNNs) emerge as a promising approach. In this study, we cast RFI detection as a supervised multi-variate time-series segmentation problem. Notably, our investigation explores the encoding of radio astronomy visibility data for SNN inference, considering six encoding schemes: rate, latency, delta-modulation, and three variations of the step-forward algorithm. We train a small two-layer fully connected SNN on simulated data derived from the Hydrogen Epoch of Reionization Array (HERA) telescope and perform extensive hyper-parameter optimization. Results reveal that latency encoding exhibits superior performance, achieving a per-pixel accuracy of 98.8% and an f1-score of 0.761. Remarkably, these metrics approach those of contemporary RFI detection algorithms, notwithstanding the simplicity and compactness of our proposed network architecture. This study underscores the potential of RFI detection as a benchmark problem for SNN researchers, emphasizing the efficacy of SNNs in addressing complex time-series segmentation tasks in radio astronomy.
△ Less
Submitted 10 June, 2024;
originally announced June 2024.
-
Formal Definition and Implementation of Reproducibility Tenets for Computational Workflows
Authors:
Nicholas J. Pritchard,
Andreas Wicenec
Abstract:
Computational workflow management systems power contemporary data-intensive sciences. The slowly resolving reproducibility crisis presents both a sobering warning and an opportunity to iterate on what science and data processing entails. The Square Kilometre Array (SKA), the world's largest radio telescope, is among the most extensive scientific projects underway and presents grand scientific coll…
▽ More
Computational workflow management systems power contemporary data-intensive sciences. The slowly resolving reproducibility crisis presents both a sobering warning and an opportunity to iterate on what science and data processing entails. The Square Kilometre Array (SKA), the world's largest radio telescope, is among the most extensive scientific projects underway and presents grand scientific collaboration and data-processing challenges. This work presents a scale and system-agnostic computational workflow model and extends five well-known reproducibility tenets into seven defined for our workflow model. Subsequent implementation of these definitions, powered by blockchain primitives, into the Data Activated Flow Graph Engine (DALiuGE), a workflow management system for the SKA, demonstrates the possibility of facilitating automatic formal verification of scientific quality in amortized constant time. We validate our approach with a simple yet representative astronomical processing task; filtering a noisy signal with a lowpass filter with both CPU and GPU methods. Our framework illuminates otherwise obscure scientific discrepancies and similarities between principally identical workflow executions.
△ Less
Submitted 3 June, 2024;
originally announced June 2024.
-
RFI Detection with Spiking Neural Networks
Authors:
Nicholas J. Pritchard,
Andreas Wicenec,
Mohammed Bennamoun,
Richard Dodson
Abstract:
Detecting and mitigating Radio Frequency Interference (RFI) is critical for enabling and maximising the scientific output of radio telescopes. The emergence of machine learning methods has led to their application in radio astronomy, and in RFI detection. Spiking Neural Networks (SNNs), inspired by biological systems, are well-suited for processing spatio-temporal data. This study introduces the f…
▽ More
Detecting and mitigating Radio Frequency Interference (RFI) is critical for enabling and maximising the scientific output of radio telescopes. The emergence of machine learning methods has led to their application in radio astronomy, and in RFI detection. Spiking Neural Networks (SNNs), inspired by biological systems, are well-suited for processing spatio-temporal data. This study introduces the first exploratory application of SNNs to an astronomical data-processing task, specifically RFI detection. We adapt the nearest-latent-neighbours (NLN) algorithm and auto-encoder architecture proposed by previous authors to SNN execution by direct ANN2SNN conversion, enabling simplified downstream RFI detection by sampling the naturally varying latent space from the internal spiking neurons. Our subsequent evaluation aims to determine whether SNNs are viable for future RFI detection schemes. We evaluate detection performance with the simulated HERA telescope and hand-labelled LOFAR observation dataset the original authors provided. We additionally evaluate detection performance with a new MeerKAT-inspired simulation dataset that provides a technical challenge for machine-learnt RFI detection methods. This dataset focuses on satellite-based RFI, an increasingly important class of RFI and is an additional contribution. Our approach remains competitive with existing methods in AUROC, AUPRC and F1 scores for the HERA dataset but exhibits difficulty in the LOFAR and Tabascal datasets. Our method maintains this accuracy while completely removing the compute and memory-intense latent sampling step found in NLN. This work demonstrates the viability of SNNs as a promising avenue for machine-learning-based RFI detection in radio telescopes by establishing a minimal performance baseline on traditional and nascent satellite-based RFI sources and is the first work to our knowledge to apply SNNs in astronomy.
△ Less
Submitted 22 March, 2024; v1 submitted 24 November, 2023;
originally announced November 2023.
-
A Bibliometric Review of Neuromorphic Computing and Spiking Neural Networks
Authors:
Nicholas J. Pritchard,
Andreas Wicenec,
Mohammed Bennamoun,
Richard Dodson
Abstract:
Neuromorphic computing and spiking neural networks aim to leverage biological inspiration to achieve greater energy efficiency and computational power beyond traditional von Neumann architectured machines. In particular, spiking neural networks hold the potential to advance artificial intelligence as the basis of third-generation neural networks. Aided by developments in memristive and compute-in-…
▽ More
Neuromorphic computing and spiking neural networks aim to leverage biological inspiration to achieve greater energy efficiency and computational power beyond traditional von Neumann architectured machines. In particular, spiking neural networks hold the potential to advance artificial intelligence as the basis of third-generation neural networks. Aided by developments in memristive and compute-in-memory technologies, neuromorphic computing hardware is transitioning from laboratory prototype devices to commercial chipsets; ushering in an era of low-power computing. As a nexus of biological, computing, and material sciences, the literature surrounding these concepts is vast, varied, and somewhat distinct from artificial neural network sources. This article uses bibliometric analysis to survey the last 22 years of literature, seeking to establish trends in publication and citation volumes (III-A); analyze impactful authors, journals and institutions (III-B); generate an introductory reading list (III-C); survey collaborations between countries, institutes and authors (III-D), and to analyze changes in research topics over the years (III-E). We analyze literature data from the Clarivate Web of Science using standard bibliometric methods. By briefly introducing the most impactful literature in this field from the last two decades, we encourage AI practitioners and researchers to look beyond contemporary technologies toward a potentially spiking future of computing.
△ Less
Submitted 13 April, 2023;
originally announced April 2023.
