Outlier Removal

Outlier removal aims to identify and mitigate the influence of data points that deviate significantly from the norm, improving the accuracy and robustness of machine learning models. Current research focuses on developing efficient and effective outlier detection methods, often integrated within model training or post-processing, and exploring the impact of outlier removal on model explainability and quantization performance across various architectures, including transformers and conformers. This is crucial for enhancing the reliability and trustworthiness of models, particularly in sensitive applications like medicine and autonomous systems, where the consequences of erroneous predictions can be severe.