The next Robust Optimization Webinar will take place this Friday, March 6, at 15:00 (CET).
Speaker: @emiliocarrizosa.bsky.social (University of Seville)
Title: Making Counterfactual Explanations Robust
More information on our webpage: sites.google.com/view/row-ser...
#ROW
From just reading their webpage for 5 minutes I would say itβs this accelerationist silicon valley mindset as βwe solve all problems with tech and AIβ
First paper and you directly chose to use CG inside another CG haha π
Come to Amsterdam!
#ismp2027
Some more paper news:
Our article "Computing weak counterfactual explanations for linear optimization: A new class of bilevel models and a tailored penalty alternating direction method" (jointly with @hlefebvr.bsky.social) has now been published in EJOR:
www.sciencedirect.com/science/arti...
The next Robust Optimization Webinar takes place this Friday, February 20, at 15:00 (CET).
Speaker 1: Justin Starreveld (University of Amsterdam)
Speaker 2: Irina Wang (Princeton University)
More information on our webpage: sites.google.com/view/row-ser...
postdoc opportunity for three years in Eindhoven, within the project "Exploiting Symmetries for Faster Bilevel Optimization Algorithms," together with Christopher Hojny #orms
www.tue.nl/en/working-a...
This is joint work with Dick den Hertog and @sibirbil.bsky.social
Our findings show that the structure of the distance function and the robustness of the counterfactual model have a significant impact on the model's privacy. In summary, non-differentiable norms and robustness increase the privacy of the model.
Second, we derive bounds on the number of queries needed to extract the model's parameters for (robust) counterfactual queries under arbitrary norm-based distances.
Applying #duality and #robust #optimization techniques, first, we derive novel mathematical formulations for the classification regions for which the decision of the unknown model is known, without recovering any of the model parameters.
Due to an increasing demand for explanations, this may involve counterfactual queries besides the typically considered factual queries.
We consider three types of queries: factual, #counterfactual, and robust counterfactual.
In our new work lead by Daan Otto we study model extraction attacks on linear machine learning models.
optimization-online.org/2026/02/line...
The goal is to reveal the parameters of a black-box machine learning model by querying the model for a selected set of data points.
The next Robust Optimization Webinar will take place this Friday, February 6, at 15:00 (CET):
Speaker: Beste Basciftci (University of Iowa)
Title: Distributionally Robust Optimization under Multimodal Decision-Dependent Uncertainty
More information on our webpage: sites.google.com/view/row-ser...
The results provide new insights into how large the K in K-Adaptability has to be chosen to provide optimal or near-optimal solutions for the two-stage robust problem. At the same time the bounds give insights on how diverse (or complex) a decision rule has to be.
Hence, in situations where the uncertainty depends only on a small number of parameters (e.g. risk factors), the number of required solutions is quite small.
The bounds I derive show that for objective uncertainty the number of solutions is essentially the dimension of the uncertainty set, while for constraint uncertainty it can be exponential in this dimension.
I am studying the question of how many different second-stage solutions are needed in two-stage #robust #optimization with integer decision variables to ensure optimality or certain approximation guarantees.
I am very proud that my recent single-authored paper on #Robust K-adaptability #Optimization got published now in Mathematical Programming:
rdcu.be/e1hsS
More information on our webpage: sites.google.com/view/row-ser...
The first Robust Optimization Webinar of this year takes place this Friday, January 23, at 15:30 (CET).
Speaker: Vineet Goyal (Columbia University)
Title: Distributionally Robust Newsvendor on a Metric
Please note the changed starting time.
Thanks to all the upcoming speakers: Vineet Goyal, Beste Basciftci, Justin Starreveld, Irina Wang, @emiliocarrizosa.bsky.social, Soroosh Shafiee, Bo Zeng, Halil Δ°brahim Bayrak, Menglei Jia, Peter Zhang, Igor Malheiros, Amir Ardestani-Jaafari
Season 7 of the Robust Optimization Webinar will start on January 23! We have an amazing list of speakers again this year! For more information please visit our
Webpage: lnkd.in/eujnAkBA
Youtube Channel: lnkd.in/eVVc9JRr
#ROW
Kicking off 2026, here is a new list of summer schools on artificial intelligence, data science, machine learning, and operations research:
thiagoserra.com/2026/01/04/s...
The US has been a violent rogue state breaking international law since decades. It is disgusting how Trump violates international law, but blaming the trump voters to have made the US a violent rogue state is a highly misleading historical amnesia.
More infos on our webpage: sites.google.com/view/row-ser...
Join us for the last ROW of this year, on Friday Dec. 12, at 17:00 (CET) (note the changed start time).
Speaker: Δ°hsan YanΔ±koΔlu (Γzyegin University)
Title: Robust Optimization under Separable Uncertainty for Electromobility and Sustainable Transportation
Finally, we develop a partition-based #branch & #bound method which is able to solve the problem for very generic two-stage problems with potentially non-linear objective and constraint functions.
In this work we derive #approximation #guarantees the K-adaptability problem achieves for the fully-adaptive two-stage problem. We also provide conditions under which the K-adaptability problem yields an optimal solution of the fully-adaptive problem.
The resulting problem can be seen as a problem over partitions of the scenario set, where each of the K second-stage solution has to "handle" a subset of scenarios.
