AI RESEARCH

Scalable Energy-Based Models via Adversarial Training: Unifying Discrimination and Generation

arXiv CS.LG

ArXi:2510.13872v4 Announce Type: replace Simultaneously achieving robust classification and high-fidelity generative modeling within a single framework presents a significant challenge. Hybrid approaches, such as Joint Energy-Based Models (JEM), interpret classifiers as EBMs but are often limited by the instability and poor sample quality inherent in