Title: Trustworthy Machine Learning: Trade-offs and Algorithms

Abstract: Ensuring the trustworthiness of machine learning systems requires balancing multiple, often conflicting, objectives such as robustness, privacy, fairness and efficiency. These trade-offs are fundamental to designing algorithms that perform reliably in adversarial or uncertain environments. This talk explores key challenges in trustworthy machine learning, highlighting theoretical limitations and algorithmic solutions. We discuss recent advances in robustness and privacy-preserving learning techniques, with a focus on their practical implications in distributed and large-scale settings.