Discrete Optimization and Machine Learning
Sebastian Pokutta (SoSe 2023 / Seminar)
Sebastian Pokutta (SoSe 2023 / Seminar)
Both Machine Learning methods as well as Discrete Optimization methods are important tools in many real-world applications. In this course we will primarily study the interplay of integer programming and, more broadly, discrete optimization methods and machine learning.The format of the course is strongly research-oriented requiring active participation of the students. The course will be a mix of traditional lectures, student presentation, discussions, and project work. In the first few weeks in-class projects will be defined.
Prerequisites: Linear Algebra, Analysis, and Discrete Optimization (ADM I/ADM II)
Registration: via email to Jannis Halbey
Office hours: by appointment
Organization & Requirements:
Students choose one of the papers below to work on. Up to 2 students can work on the same paper. Students are expected to individually
Write a report (5 page minimum in Latex) about the paper and the contribution from the students themselves.
During the semester, present their plans for the final report and presentation in a shorter 5-minute presentation.
Give a final presentation to present their findings (details discussed in the first meeting).
Timeline:
18.04.2023, 11:00 a.m. (Berlin time) in Room MA 651: Seminar outline
02.05.2023, 11:59 p.m. (Berlin time): Paper selection deadline (send your desired paper to Jannis Halbey. Assignment is first come, first served. Papers that are no longer available are crossed out.)
22.05.2023, 10:00 a.m. (Berlin time): 5-minute presentations in room MAR 0.016 (Marchstraße 23)
28.06.2023, 11:59 p.m. (Berlin time): Report submission deadline (send your reports to Jannis Halbey)
29.06.2023, 10:00 a.m. (Berlin time): Final presentations in room H 6124
30.06.2023, 10:00 a.m. (Berlin time): Final presentations in room H 3012
Nonsmooth Optimization and Machine Learning (0/2)
Nonsmooth Implicit Differentiation for Machine-Learning and Optimization
https://proceedings.neurips.cc/paper/2021/hash/70afbf2259b4449d8ae1429e054df1b1-Abstract.html
Computer Vision (2/2)
SAIF: Sparse Adversarial and Interpretable Attack Framework.
https://arxiv.org/pdf/2212.07495.pdf
Feature transformation / extraction (2/2)
Vanishing Component Analysis
https://proceedings.mlr.press/v28/livni13.html
Optimization and Machine Learning (2/2)
Linear coupling: An ultimate unification of gradient and mirror descent
https://arxiv.org/abs/1407.1537
Deep Ensembling (1/2)
PEP: Parameter Ensembling by Perturbation
https://proceedings.neurips.cc/paper/2020/file/652c208b21f13f6e995bfc1154a1a2e5-Paper.pdf
Deep (Hybrid) Learning (2/2)
Deep Hybrid Models: Bridging Discriminative and Generative Approaches
https://www.cs.cornell.edu/~kuleshov/papers/uai2017.pdf
Optimization, Federated Learning, Adversarial Robustness (0/2)
Byzantine Stochastic Gradient Descent
https://arxiv.org/abs/1803.08917
Machine Learning for column generation (2/2)
COIL: A Deep Architecture for Column Generation
https://optimization-online.org/2022/09/coil-a-deep-architecture-for-column-generation/
(Practical) graph isomorphisms (0/2)
The complexity of McKay’s canonical labelling algorithm
https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=f6ea1e075e2edd4975504781142a0fd8ee71383d
Convex constrained optimization (0/2)
Frank-Wolfe with a nearest extreme point oracle
http://proceedings.mlr.press/v134/garber21a/garber21a.pdf
Optimization (2/2)
Scheduling electric vehicles
https://link.springer.com/article/10.1007/s12469-017-0164-0
Data-driven optimization (0/2)
Data-Driven Approximation of the Perron-Frobenius Operator Using the Wasserstein Metric
https://www.sciencedirect.com/science/article/pii/S2405896322027094
Convex optimization, linear minimization (0/2)
Heavy Ball Momentum for Conditional Gradient
https://arxiv.org/abs/2110.04243
Robust Combinatorial Optimization (0/2)
A branch and bound algorithm for robust binary optimization with budget uncertainty,
https://link.springer.com/article/10.1007/s12532-022-00232-2
RKHS and dynamical Systems (0/2)
Embedding classical dynamics in a quantum computer
https://arxiv.org/pdf/2012.06097.pdf