Quasars experiencing strong lensing offer unique viewpoints on subjects like
the cosmic expansion rate, the dark matter profile within the foreground
deflectors, and the quasar host galaxies. Unfortunately, identifying them in
astronomical images is challenging since they are overwhelmed by the abundance
of non-lenses. To address this, we have developed a novel approach by
ensembling cutting-edge convolutional networks (CNNs) -- i.e., ResNet,
Inception, NASNet, MobileNet, EfficientNet, and RegNet -- along with vision
transformers (ViTs) trained on realistic galaxy-quasar lens simulations based
on the Hyper Suprime-Cam (HSC) multiband images. While the individual model
exhibits remarkable performance when evaluated against the test dataset,
achieving an area under the receiver operating characteristic curve of $>$97.4%
and a median false positive rate of 3.1%, it struggles to generalize in real
data, indicated by numerous spurious sources picked by each classifier. A
significant improvement is achieved by averaging these CNNs and ViTs, resulting
in the impurities being downsized by factors up to 40. Subsequently, combining
the HSC images with the UKIRT, VISTA, and unWISE data, we retrieve
approximately 60 million sources as parent samples and reduce this to 892,609
after employing a photometry preselection to discover $z>1.5$ lensed quasars
with Einstein radii of $\theta_\mathrm{E}<5$ arcsec. Afterward, the ensemble
classifier indicates 3991 sources with a high probability of being lenses, for
which we visually inspect, yielding 161 prevailing candidates awaiting
spectroscopic confirmation. These outcomes suggest that automated deep learning
pipelines hold great potential in effectively detecting strong lenses in vast
datasets with minimal manual visual inspection involved.

本研究使用卷积神经网络和深度学习管道自动检测强透镜效应，并在多波段图像中检测到了可以用于光谱学确认的 161 个候选源。

通过组合神经网络简化多波段映射中的透镜状类星体鉴定

Streamlined Lensed Quasar Identification in Multiband Images via  Ensemble Networks

This is the fourth paper in a series of publications aiming at discovering
quasars at the epoch of reionization. In this paper, we expand our search for
$z\sim 7$ quasars to the footprint of the Dark Energy Survey (DES) Data Release
One (DR1), covering $\sim 5000$ deg$^2$ of new area. We select $z\sim 7$ quasar
candidates using deep optical, near-infrared (near-IR) and mid-IR photometric
data from the DES DR1, the VISTA Hemisphere Survey (VHS), the VISTA Kilo-degree
Infrared Galaxy (VIKING) survey, the UKIRT InfraRed Deep Sky Surveys -- Large
Area Survey (ULAS) and the unblurred coadds from the Wide-field Infrared Survey
Explore ($WISE$) images (unWISE). The inclusion of DES and unWISE photometry
allows the search to reach $\sim$ 1 magnitude fainter, comparing to our $z
\gtrsim 6.5$ quasar survey in the northern sky (Wang et al. 2018). We report
the initial discovery and spectroscopic confirmation of six new luminous
quasars at $z>6.4$, including an object at $z=7.02$, the fourth quasar yet
known at $z>7$, from a small fraction of candidates observed thus far. Based on
the recent measurement of $z \sim 6.7 $ quasar luminosity function using the
quasar sample from our survey in the northern sky, we estimate that there will
be $\gtrsim$ 55 quasars at $z > 6.5$ at $M_{1450} < -24.5$ in the full DES
footprint.

在该文章中，作者使用来自多源的光学、近红外和中红外光学深度数据搜索赤化时期的类星体，发现了六个新的亮类星体及一个具有 $z=7.02$ 的亮且红的新类星体，三次观测和发现了这些候选对象。