我们组希望以概率论和统计学习为基础,来建立高度实用、可靠的机器学习新方法。我们专注于概率机器学习 (Probabilistic machine learning) 和贝叶斯深度学习(Bayesian deep learning),致力于开发具有理论保证的快速、稳健的训练和推断方法,以及探索它们在真实大数据与深度神经网络上的应用。我们的研究方向包括但不限于:
1. Reliable and scalable probabilistic modeling (such as uncertainty quantification, Bayesian inference, MCMC)
2. Large-scale training & inference (such as stochastic algorithms, quantization, sparse learning)
3. Learning with discrete structure (such as graph, text, binary neural networks)
4. Learning under privacy, fairness, safety constraints
1. 计算机,统计,数学等相关专业本科或本科以上学历;
2. 具备概率,统计,机器学习基本知识;
3. 熟悉Pytorch, Jax等深度学习框架。