Depression is the most common mental illness that affects both the mental and physical health of the human body. Depression often leads the sufferers to commit suicide. Consequently, it is necessary to detect the depression state of the suffering. Depression detection in social media is designed to detect depressed users and analyze their depression level via the information shared on social media. Compared to machine learning, deep learning models are preferably used for depression detection in social media due to their unstructured data handling capability. However, deep learning models are solely dependent on a certain domain which are insufficient to detect the depression state of the user accurately.
   Cross-domain is the collective approach that manipulates feedback from multiple domains to improve the performance of the model. Associating different domain information of the users in the social media helps better detect depression state. Cross-domain for depression detection transfers the relevant information of the suffer across heterogeneous domains via social media and improves the performance of the depression detection model.