研究目的
To present a new fully-automatic classification model to select extragalactic objects within astronomical photometric catalogs.
研究成果
The study successfully demonstrates the application of automated feature selection, anomaly detection, and classification in the task of star-galaxy separation, identifying about 38 million extragalactic objects in the WISExPan-STARRS1 data.
研究不足
The study does not address the potential variability in the quality of photometric data across different surveys and its impact on classification accuracy.
1：Experimental Design and Method Selection:
The methodology involves using an autoencoder neural network for feature space construction and Support Vector Machine (SVM) for classification.
2：Sample Selection and Data Sources:
The training sample consists of
3：7 million objects from the SDSS DR14 catalog, cross-matched with WISE and Pan-STARRS1 catalogs. List of Experimental Equipment and Materials:
The study utilizes data from the WISE and Pan-STARRS1 surveys.
4：Experimental Procedures and Operational Workflow:
The process includes data representation, feature space construction, anomaly detection, and classification.
5：Data Analysis Methods:
The analysis involves using One-Class SVM for anomaly detection and kernel-SVM for classification.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
