Pulsar Detection
Multi-Model Machine Learning Architecture for Automated Pulsar Detection
Easy to use Interface for Detecting Pulsars with the overall Accuracy of 97.8% and Recall of 95.3%
Pulsars are rapidly rotating neutron stars that blast out pulses of radiation at regular intervals ranging from seconds to milliseconds.
Read More about PulsarsWe aggregated numerical and image data from .phcx files utilizing PDMP and PEASOUP. Numerical attributes, such as signal-to-noise ratio and dispersion measure, are processed through an Artificial Neural Network (ANN), while image data—representing visualizations of pulsar signals—is fed into a Convolutional Neural Network (CNN). The ANN leverages features like Profile_mean and DM_skewness to generate probabilistic scores, while the CNN extracts spatial hierarchies from the image channels. These probabilistic outputs are then ensembled, synthesizing a comprehensive inference model that enhances the robustness and accuracy of pulsar detection by integrating both numerical and visual modalities to acquire Multi-Model Prowess.
Detect Pulsars Now