← Search

Tony T. Wang

2 accepted papers

2026

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

ICLR 2026poster

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…

Cited by 0SourcecodeScholar