Events organized

Non-convex optimization has become increasingly important for modern data science applications, with many unsolved challenging open problems. Due to the absence of convexity, the well-developed paradigms of convex optimization cannot be fully extended here, leading to the current situation that the theory of non-convex optimization is way behind the practice. In this mini-symposium, we focus on recent advances in non-convex optimization, including the analysis and understanding of the fundamental nature of the non-convex optimization, algorithmic bias/implicit regularization of gradient-based algorithms, fast convergent algorithms in non-convex problems, and their applications in inverse problems, imaging and machine learning and many others.

随着科学计算中数据量的不断增加和数学建模精准性要求的不断提 高，算法效率逐渐成为实际应用问题求解时的瓶颈，对快速算法的 需求日益增加。本论坛将聚焦近几年算法加速技巧的发展，包含惯 性加速、随机算法等方面，也同时关注他们在实际问题中的应用。 计划邀请应用与算法两方面的专家进行交流，探讨更加有效的实际问题求解算法等发展方向。