Quantum-Classical Deep Learning Models for Image Classification. The Search for Extraterrestrial Intelligence Institute (SETI) was established to gather and analyze radio signals from space in an effort to detect signs of advanced civilizations that may possibly exist elsewhere in the cosmos. A two-dimensional spectrogram is created using time-series raw data obtained from the Allen telescope. We suggest the use of a Hybrid Quantum-Classical convolutional Neural network (HQCNN) using a Quantum node serving to extract intricate features of SETI radio images that are being classified with classical convolutional Neural Network. We also generalized the combination of quantum gates in the Quantum node to improvise the performance of HQCNN. Quantum superposition and entanglement properties of Quantum computers render it feasible to comprehend the basic relationships in the data set more precisely than with traditional Machine Learning and Deep Learning techniques. In this context, with the notion of leveraging the computational benefit of an HQCNN, we, therefore, treated this challenging task as binary classification and filtered any two classes from the primary data set. The experimental analysis was executed in the Google Collaboratory platform with the Python packages such as Pennylane, Keras, and TensorFlow. To validate the outcomes, the HQCNN model is contrasted with various classical Deep learning models in terms of diverse performance metrics. The proposed model outperforms the other models in categorizing Allen Telescope data with the highest accuracy of 90.19% while utilizing a very small portion of the original data set, which motivates further research into the use of Quantum Machine Learning algorithms for signal processing tasks.
--> The data is available on the official SETI website. The four files have seperate train and test files.
--> All the models have been provided in the "Models" folder. It also contains the preprocessing and encoding part.