Abstract:Spiking Neural Networks (SNNs) promise efficient spatio-temporal data processing owing to their dynamic nature. This paper addresses a significant challenge in radio astronomy, Radio Frequency Interference (RFI) detection, by reformulating it as a time-series segmentation task inherently suited for SNN execution. Automated RFI detection systems capable of real-time operation with minimal energy consumption are increasingly important in modern radio telescopes. We explore several spectrogram-to-spike encoding methods and network parameters, applying first-order leaky integrate-and-fire SNNs to tackle RFI detection. To enhance the contrast between RFI and background information, we introduce a divisive normalisation-inspired pre-processing step, which improves detection performance across multiple encoding strategies. Our approach achieves competitive performance on a synthetic dataset and compelling results on real data from the Low-Frequency Array (LOFAR) instrument. To our knowledge, this work is the first to train SNNs on real radio astronomy data successfully. These findings highlight the potential of SNNs for performing complex time-series tasks, paving the way for efficient, real-time processing in radio astronomy and other data-intensive fields.

What problem does this paper attempt to address?

This paper attempts to solve a key problem in radio astronomy, namely **Radio - Frequency Interference (RFI) detection**. Specifically, the author proposes a method based on Spiking Neural Networks (SNNs), redefining RFI detection as a time - series segmentation task to achieve real - time, low - power RFI detection. ### Main problems and challenges 1. **The impact of RFI on observational data**: Radio - Frequency Interference (RFI) refers to radio - wave signals from man - made or terrestrial sources, which can severely contaminate radio - astronomical observational data. Traditional RFI detection methods usually need to complete the entire observation before processing, which not only increases the storage cost but also reduces the data - processing efficiency. 2. **Real - time and energy - consumption requirements**: Modern radio telescopes have increasingly high requirements for RFI detection systems, especially the need for real - time operation capabilities and low - energy - consumption characteristics. Although existing machine - learning methods are effective, they often require a large amount of training data and high - computing resources and are difficult to meet the needs of real - time processing. 3. **Characteristics of time - series data**: Radio - astronomy data has obvious time - series characteristics, and traditional two - dimensional image - processing methods cannot fully utilize this characteristic. Therefore, how to use the time - dynamic characteristics of SNNs to process such data is an important research direction. ### Solutions The author has solved the above problems through the following methods: 1. **Redefining RFI detection as a time - series segmentation task**: The author uses the time - dynamic characteristics of SNNs to regard RFI detection as a time - series segmentation task rather than a traditional two - dimensional image - segmentation task. This method can better capture the time - dependence of data and is more suitable for real - time processing. 2. **Introducing multiple encoding methods**: In order to convert spectrograms into pulse sequences that can be processed by SNNs, the author has explored multiple encoding methods, including Latency, Rate, Delta - Modulation, Delta - Exposure, and Step - Forward. These encoding methods can help SNNs better understand the input data. 3. **Pre - processing techniques**: The author has introduced a pre - processing step inspired by Divisive Normalisation to enhance the contrast between RFI and background information. This method significantly improves the detection performance under different encoding strategies. 4. **Experimental verification**: The author has carried out experiments on synthetic data sets (simulating the Hydrogen Epoch of Reionisation Array, HERA) and real data sets (Low - Frequency Array, LOFAR) to verify the effectiveness of the proposed method. The results show that the SNN using Latency encoding and Divisive Normalisation performs excellently in multiple indicators, approaching or even exceeding the performance of existing traditional algorithms and ANN methods. ### Conclusion This paper shows the potential of SNNs in processing complex time - series tasks such as RFI detection, especially having unique advantages in real - time processing and low - energy consumption. By introducing new encoding methods and pre - processing techniques, the author has successfully improved the performance of SNNs in radio - astronomy data processing, providing new ideas for further research and development of efficient RFI detection systems.

Spiking Neural Networks for Radio Frequency Interference Detection in Radio Astronomy

Supervised Radio Frequency Interference Detection with SNNs

RFI Detection with Spiking Neural Networks

RSNN: Recurrent Spiking Neural Networks for Dynamic Spatial-Temporal Information Processing

A Self-learning Neural Network Approach for Radio Frequency Interference Detection and Removal in Radio Astronomy

Semantic Segmentation of Solar Radio Spikes at Low Frequencies

Spike-based Encoding and Learning of Spectrum Features for Robust Sound Recognition.

Spiking Neural Networks for Detecting Satellite-Based Internet-of-Things Signals

Learned Spike Encoding of the Channel Response for Low-Power Environment Sensing

Spiking Neural Network for Fourier Transform and Object Detection for Automotive Radar

Bipolar Population Threshold Encoding for Audio Recognition with Deep Spiking Neural Networks

Event-driven Spiking Neural Networks with Spike-Based Learning

SpikingNeRF: Making Bio-inspired Neural Networks See through the Real World

A CNN and LSTM-Based Approach to Classifying Transient Radio Frequency Interference

A Spiking Neural Network Model for Sound Recognition.

Spike encoding techniques for IoT time-varying signals benchmarked on a neuromorphic classification task

SPIDEN: deep Spiking Neural Networks for efficient image denoising

Learning to detect RFI in radio astronomy without seeing it

A robust RFI identification for radio interferometry based on a convolutional neural network

A frugal Spiking Neural Network for unsupervised classification of continuous multivariate temporal data