I
Ilija Bogunovic
Researcher at ETH Zurich
Publications - 53
Citations - 1073
Ilija Bogunovic is an academic researcher from ETH Zurich. The author has contributed to research in topics: Submodular set function & Bayesian optimization. The author has an hindex of 15, co-authored 47 publications receiving 801 citations. Previous affiliations of Ilija Bogunovic include École Polytechnique Fédérale de Lausanne.
Papers
More filters
Posted Content
Near-Optimally Teaching the Crowd to Classify
TL;DR: In this paper, the authors propose a natural stochastic model of the learners, modeling them as randomly switching among hypotheses based on observed feedback, and develop an efficient algorithm for selecting examples to teach to workers.
Journal ArticleDOI
Learning-Based Compressive Subsampling
TL;DR: This paper formulate combinatorial optimization problems seeking to maximize the energy captured in these signals in an average-case or worst-case sense, and shows that these can be efficiently solved either exactly or approximately via the identification of modularity and submodularity structures.
Proceedings Article
Near-Optimally Teaching the Crowd to Classify
TL;DR: This work proposes a natural stochastic model of the learners, modeling them as randomly switching among hypotheses based on observed feedback, and develops STRICT, an efficient algorithm for selecting examples to teach to workers.
Posted Content
High-Dimensional Bayesian Optimization via Additive Models with Overlapping Groups
TL;DR: In this article, a message passing algorithm is proposed to optimize the acquisition function in high-dimensional black-box functions. But the authors consider the assumption that the subsets are disjoint, and consider additive models with arbitrary overlap among the sub-sets.
Proceedings Article
Adversarially Robust Optimization with Gaussian Processes
TL;DR: In this article, the authors consider the problem of Gaussian process optimization with an added robustness requirement: the returned point may be perturbed by an adversary, and they require the function value to remain as high as possible even after this perturbation.