In the paper 'Addressing the Loss-Metric Mismatch with Adaptive Loss Alignment', the authors propose a sample-efficient reinforcement learning approach for adapting the loss function dynamically during training to directly optimize the evaluation metric.

Authors

• Person: Atharva Kulkarni, Yuan Zhang, Joel Ruben Antony Moniz, Xiou Ge, Bo-Hsiang Tseng, Dhivya Piraviperumal, Swabha Swayamdipta, Hong Yu Organization: Apple Machine Learning Research
  Evaluating Evaluation Metrics — The Mirage of Hallucination ...

Sources

• Evaluating Evaluation Metrics — The Mirage of Hallucination ... machinelearning.apple.com Atharva Kulkarni, Yuan Zhang, Joel Ruben Antony Moniz, Xiou Ge, Bo-Hsiang Tseng, Dhivya Piraviperumal, Swabha Swayamdipta, Hong Yu · Apple Machine Learning Research via serper

Referenced by nodes (2)

• reinforcement learning concept
• evaluation metrics concept