Navigation
  • About
  • People
  • Publications
  • Blog
  • Teaching
  • Student Projects
  • Openings
  • Contact
  • About
  • People
  • Publications
  • Blog
  • Teaching
  • Student Projects
  • Openings
  • Contact

Probabilistic Modeling and Inference

A central challenge in machine learning is to quantify uncertainty via probabilistic models that capture statistical dependencies between many uncertainty quantities. We have recently initiated the study of learning and inference in Probabilistic Submodular Models, a rich class of probabilistic models defined through submodular functions. This class contains and generalizes a number of extensively studied subclasses, such as determinantal point processes, and Ising models. A major benefit of such models is that they allow to capture complex, long-range interactions between many variables, which is useful, for example in computer vision and information retrieval. We develop novel algorithms for efficient approximate inference, using, e.g., variational or sampling techniques, as well as learning such models from data.

Publications

2018
  • Discrete Sampling using Semigradient-based Product Mixtures
  • A. Gotovos, H. Hassani, A. Krause, S. Jegelka
  • In Conference on Uncertainty in Artificial Intelligence (UAI), 2018
  • Oral presentation
  • [bibtex] [abstract] [pdf] [long] [talk] [poster]
2017
  • Differentiable Learning of Submodular Models
  • J. Djolonga, A. Krause
  • In Neural Information Processing Systems (NeurIPS), 2017
  • Spotlight presentation
  • [bibtex] [abstract] [pdf] [code]
  • Improving Optimization-Based Approximate Inference by Clamping Variables
  • J. Zhao, J. Djolonga, S. Tschiatschek, A. Krause
  • In Conference on Uncertainty in Artificial Intelligence (UAI), 2017
  • [bibtex] [abstract] [pdf]
2016
  • Variational Inference in Mixed Probabilistic Submodular Models
  • J. Djolonga, S. Tschiatschek, A. Krause
  • In Proc. Neural Information Processing Systems (NeurIPS), 2016
  • [bibtex] [abstract] [pdf] [video]
  • Cooperative Graphical Models
  • J. Djolonga, S. Jegelka, S. Tschiatschek, A. Krause
  • In Proc. Neural Information Processing Systems (NeurIPS), 2016
  • [bibtex] [abstract] [pdf] [video]
  • Learning Probabilistic Submodular Diversity Models Via Noise Contrastive Estimation
  • S. Tschiatschek, J. Djolonga, A. Krause
  • In Proc. International Conference on Artificial Intelligence and Statistics (AISTATS), 2016
  • [bibtex] [abstract] [pdf]
2015
  • Sampling from Probabilistic Submodular Models
  • A. Gotovos, S. H. Hassani, A. Krause
  • In Proc. Neural Information Processing Systems (NeurIPS), 2015
  • Oral presentation
  • [bibtex] [abstract] [pdf] [long] [talk] [poster] [video]
  • Higher-Order Inference for Multi-class Log-supermodular Models
  • J. Zhang, J. Djolonga, A. Krause
  • In International Conference on Computer Vision (ICCV), 2015
  • [bibtex] [abstract] [pdf] [long]
  • Scalable Variational Inference in Log-supermodular Models
  • J. Djolonga, A. Krause
  • In International Conference on Machine Learning (ICML), 2015
  • [bibtex] [abstract] [pdf] [long]
2014
  • From MAP to Marginals: Variational Inference in Bayesian Submodular Models
  • J. Djolonga, A. Krause
  • In Proc. Neural Information Processing Systems (NeurIPS), 2014
  • [bibtex] [abstract] [pdf] [long]
  • About
  • People
  • Publications
  • Blog
  • Teaching
  • Student Projects
  • Openings
  • Contact

Learning & Adaptive Systems Group | Institute for Machine Learning | © 2025 ETH Zurich