The measurement and analysis of brain activity are considered essential aspects of neuroscience. Worldwide, schizophrenia (Scz) is a brain disorder that affects people’s feelings, thoughts, and behaviors. Therefore, an accurate and quick detection approach is required for the patients to receive the proper care and high-quality treatment. Because it captures brain activity, electroencephalography (EEG) is an effective biomarker for Scz detection. Several deep learning and machine learning techniques have been created, yet they cannot produce good results. Hence, an automated process is required to accurately classify the EEG signals between schizophrenia and healthy control signals. To overcome this issue, this paper aims to enhance the effectiveness of EEG-based Scz detection. Therefore, we proposed an EEG-based solution for diagnosing schizophrenia using an autoencoder model. The raw EEG signals are passed as the input to the encoder part, which extracts the most latent feature, which is further classified into schizophrenia and healthy controls. The proposed model gives a high-performance accuracy of 99.75% with the reduced numbers of the parameters and reduced computational times. The model yielded the following results: 99.75% Precision, F1 score of 99.75%, Sensitivity of 99.49%, and specificity of 100%. Furthermore, compared to techniques in the existing literature, this study’s classification performance for the diagnosis of schizophrenia was satisfactorily high.