Early-fusion based pulsar identification with smart under-sampling
نویسندگان
چکیده
The discovery of pulsars is great significance in the field physics and astronomy. As astronomical equipment produces a large amount pulsar data, an algorithm for automatically identifying becomes urgent. We propose deep learning framework recognition. In response to extreme imbalance between positive negative examples hard sample issue presented HTRU Medlat Training Data,there are two coping strategies our framework: smart under-sampling improved loss function. also apply early-fusion strategy integrate features obtained from different attributes before classification improve performance. To best knowledge,this first study that integrates these techniques together experiment results show outperforms previous works with respect either training time or F1 score. can not only speed up by 10X compared state-of-the-art work, but get competitive result terms
منابع مشابه
Sensor Fusion-Based Middleware for Smart Homes
Smart homes are sensor-rich environments that contain dynamic sets of interacting components. These components often use competing and closed standards and form a message-based architecture. This complicates the development of applications that require information from disparate sources. It becomes difficult to add new components or to allow components from different applications to interact wi...
متن کاملPulsar Candidate Identification with Artificial Intelligence Techniques
Discovering pulsars is a significant and meaningful research topic in the field of radio astronomy. With the advent of astronomical instruments such as he Five-hundredmeter Aperture Spherical Telescope (FAST) in China, data volumes and data rates are exponentially growing. This fact necessitates a focus on artificial intelligence (AI) technologies that can perform the automatic pulsar candidate...
متن کاملMulti-feature Fusion Face Recognition Based on Kernel Discriminate Local Preserve Projection Algorithm under Smart Environment
In this paper, a new face recognition method based on kernel discriminate local preserve projection(KDLPP) and Multi-feature fusion under smart environment is proposed . In order to solve the small sample size problem, combined with kernel theory and QR decomposition, a new face recognition algorithm named kernel discriminate local preserve projection is proposed based on discriminate local pre...
متن کاملSmart Sampling without Reinforcement
We conducted an experiment to compare the effect of natural frequencies and active, trial-by-trial sampling on participants’ estimation of proportions in a population. The two information formats evoked similar degrees of accuracy, sensitivity to subtle statistical properties of the environment, and sensitivity to the estimates’ accuracy. However, in some situations participants who didn’t have...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Research in Astronomy and Astrophysics
سال: 2021
ISSN: ['1674-4527', '2397-6209']
DOI: https://doi.org/10.1088/1674-4527/21/10/257