The field of generalized category discovery is moving towards more unified and unbiased approaches, with a focus on addressing the challenges of class imbalance and label bias. Recent research has introduced new frameworks and methods that jointly model old and new classes, and employ techniques such as debiased learning, distribution guidance, and probabilistic graphical models to improve the accuracy and robustness of category discovery. These innovative approaches have shown state-of-the-art performance on various benchmarks and datasets, and have the potential to be widely applied in tasks such as image classification, anomaly detection, and financial fraud detection. Notable papers include ProtoGCD, which introduces a unified and unbiased prototype learning framework, and DebGCD, which proposes a debiased learning framework with distribution guidance. Additionally, LCGC presents a debiasing scheme that encourages biased class predictions during training, and Addressing Class Imbalance with Probabilistic Graphical Models and Variational Inference proposes a method for imbalanced data classification using deep probabilistic graphical models and variational inference.