Deep Learning for Bias Detection: From Inception to Deployment

To create a more inclusive workplace, enterprises are actively investing in identifying and eliminating unconscious bias (e.g., gender, race, age, disability, elitism religion) across their various functions. We propose deep learning model with transfer based language to learn from manually tagged documents for automatically enterprise content. first pretrain learning-based language-model using Wikipedia, then fine tune the large unlabelled data set related types of Finally, linear layer followed by softmax is added at end trained on labelled dataset consisting The thoroughly evaluated independent datasets ensure general application. present proposed method its deployment detail real-world