NIPS 2017 - presentations from the Algorithms session

• Diffusion Approximations for Online Principal Component Estimation and Global Convergence • Positive-Unlabeled Learning with Non-Negative Risk Estimator • An Applied Algorithmic Foundation for Hierarchical Clustering • Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results • QSGD: Communication-Efficient SGD via Gradient Quantization and Encoding • Inhomogeneous Hypergraph Clustering with Applications • K-Medoids for K-Means Seeding • Online Learning with Transductive Regret • Matrix Norm Estimation from a Few Entries Semisupervised Clustering, AND-Queries and Locally Encodable Source Coding