Abstract:Telescopes such as the Large Sky Area Multi-Object Spectroscopic Telescope and the Sloan Digital Sky Survey have produced an extensive collection of spectra, challenging the feasibility of manual classification in terms of accuracy and efficiency. To overcome these limitations, machine learning techniques are increasingly being utilized for automated spectral classification. However, these approaches primarily treat spectra as frequency domain signals, and lack robustness in low signal-to-noise ratio (S/N) scenarios and for small datasets of rare celestial objects. Moreover, they frequently neglect nuanced expert astronomical understanding. In this study, we draw inspiration from the human spectral discrimination process and propose a new model called the Image-EFficientNetV2-Spectrum Convolutional Neural Network (IEF-SCNN). IEF-SCNN combines spectral images using EfficientNetV2 with one-dimensional (1D) spectra through a 1DCNN. This integration effectively incorporates astronomical expertise into the classification process. Specifically, we plot the spectrum as an image and then classify it in a way that incorporates an attention mechanism. This attention mechanism mimics human observation of images for classification, selectively emphasizing relevant information while ignoring irrelevant details. Experimental data demonstrate that IEF-SCNN outperforms existing models in terms of the F1-score and accuracy metrics, particularly for low S/N (<6) data. Using progressive learning and an attention mechanism, the model trained on 12,000 M-class stars with an S/N below 6 achieved an accuracy of 87.38% on a 4000-sample test set. This surpasses traditional models (support vector machine with 83.15% accuracy, random forest with 65.40%, and artificial neural network with 84.40%) and the 1D stellar spectral CNN (85.65% accuracy). This research offers a foundation for the development of innovative methods for the automated identification of specific celestial objects, and can promote the creation of user-friendly software for astronomers who may not have computational expertise.

A New Automated Spectral Feature Extraction Method and Its Application in Spectral Classification and Defective Spectra Recovery

Spectral Feature Extraction Using Partial and General Method

A Method for Auto-Extraction of Spectral Lines Based on Sparse Representation

[A Novel Spectrum Feature Extraction Method].

Spectral Classification and Particular Spectra Identification Based on Data Mining

Automatic Classification of Spectra with IEF-SCNN

DeepSpectra: an End-to-end Deep Learning Approach for Quantitative Spectral Analysis

Feature Extracting Methods in Spectrum Data Mining

Spectral Analysis and Recognition Using Multi-scale Features and Neural Networks

RVM Supervised Feature Extraction and Seyfert Spectra Classification

Deep Neural Networks with Local Connectivity and Its Application to Astronomical Spectral Data.

Spectral and Multi-Spatial-Feature Based Deep Learning for Hyperspectral Remote Sensing Image Classification.

Extracting Features by Interpolating and Down-Sampling for Galaxy and QSO Spectrum Classification

Using Deep Learning Algorithms to Perform Accurate Spectral Classification

Spectral-Spatial Feature Extraction and Classification by ANN Supervised with Center Loss in Hyperspectral Imagery

Automatic Classification Method for Astronomical Spectra Based on Wavelet Packet and SVM

Spatial-Spectral Feature Extraction of Hyperspectral Image Based on Deep Learning

Classification of Astronomical Spectra Based on Multiscale Partial Convolution

Hyperspectral Image Classification Based on Spectral Feature Extraction

Feature Extraction from Spectroscopy Using LASSO and Net Analyte Signal

Spectral–Spatial Classification of Hyperspectral Data Based on Deep Belief Network