Zero shot image classification system using an optimized generalized adversarial network

Image attribute classification is the hottest topic in the digital visualization industry. But, predicting the unseen class attributes using neural approaches is very complicated. Hence, zero-shot learning has been introduced along with the neural models. Still, there are issues with classifying the unseen class features because of poor prediction and noisy data. So, the present research article has aimed to design a novel Ant Lion-based Generalized Adversarial Intelligent Network (AL-GAIN) for attributes forecasting from unseen data. Primarily, the database has been filtered in the pre-processing phase. The error-free data is entered into the classification phase to identify and store the present attributes in the trained data. Moreover, the test data was imported, and features were extracted by the novel AL-GAIN. The similarity matching process was performed to find the unseen class attributes. The planned model has been executed in the python environment. Finally, the prediction accuracy has been measured for both seen and unseen data compared with other models and has gained better attributes in forecasting outcomes.