PhD School: Machine Learning and Optimization

Register here

About the PhD School

As part of the CWI Research Semester on Learning Enhanced Optimization, we are excited to host an International PhD School during the first week of September. The PhD School explores recent theoretical developments at the intersection of Machine Learning and Optimization.

As computation increasingly moves into real-time, data-driven, and uncertain settings, classic algorithmic paradigms are being reimagined through the lenses of learning theory, incentive design, and machine-learned predictions. The PhD School delves into the theoretical and algorithmic foundations of integrating such novel techniques into algorithm design.

This three-day event, running from Tuesday to Thursday, is tailored for PhD students interested in the growing intersection of Machine Learning and Optimization. The school combines lectures with collaborative group works.

The school will feature lectures from three renowned researchers:

• Michal Feldman (Tel Aviv University, Israel)
• Anupam Gupta (New York University, USA)
• Ola Svensson (EPFL, Switzerland)

Who should attend?

The school is aimed at PhD students in theoretical computer science, algorithms, machine learning, operations research and related areas. Please note that there will be a strong focus on theoretical and algorithmic foundations. Advanced Master’s students and early-career researchers/postdocs with a strong theoretical background are also warmly encouraged to participate.

Registration

Registration is now open.

Note that to facilitate catering during the event, a registration fee of €60 per participant is required; this fee covers lunches and refreshments for the duration of the school.

Please register here.

The registration deadline is 24 August 2025.

Please note that registrations will be processed on a first-come, first-served basis and will close once full capacity is reached.

Lectures and Schedule

Format

Each day of the PhD School consists of two parts:

• Morning sessions: The invited lecturer of the day will deliver two in-depth lectures, introducing key concepts and recent developments. At the end of the lectures, the lecturer will provide some assignment questions related to the day’s topic.
• Afternoon sessions: Participants will work collaboratively in small groups to tackle the given assignment questions and solve them. The day will conclude with a joint discussion session, where the solutions will be discussed with the lecturer. 

This format is designed to promote both deep understanding and active engagement with the material.

Day 1: Tuesday, 2 September 2025

Lecturer: Anupam Gupta, New York University

Title: Learning for Online Algorithms: The Multiplicative Weights Method

Abstract:
In these lectures, we will explore the use of the multiplicative weights method (and other connections to online learning) to online algorithm design. We will see some different ways in which a collection of simple-looking ideas can be used to solve linear programs, and give near-optimal algorithms for problems in covering and load-balancing.

Day 2: Wednesday, 3 September 2025

Lecturer: Michal Feldman, Tel Aviv University

Title: Algorithmic Contract Design

Abstract:
Contracts are a fundamental tool for shaping incentives, used by a principal to motivate agents to exert effort by linking payments to observable outcomes. As real-world contract scenarios increasingly move online, scale up, and rely on data, classic economic models face new challenges and opportunities. This lecture series introduces the emerging field of algorithmic contract theory, where tools from computer science are used to design and analyse contracts in complex and data-rich environments.

Day 3: Thursday, 4 September 2025

Lecturer: Ola Svensson, EPFL

Title: Algorithms with Predictions for Faster Execution and Better Decisions

Abstract:
In these lectures, we will explore data-driven algorithm design, focusing specifically on the theory of learning-augmented algorithms that utilize (potentially unreliable) machine-learned predictions. This rapidly growing area pushes beyond worst-case analysis. Our lectures will concentrate on how to effectively employ these predictions in the design of faster algorithms and for improving the performance of online algorithms.

Accommodation and Venue

Please be aware that hotel prices in Amsterdam can be quite steep. We strongly recommend all participants to secure their hotel reservations as early as possible!

Hotel Recommendations

• Hotel Casa
• MEININGER Hotel (Amsterdam Amstel)
• The Manor Amsterdam
• The Lancaster Hotel Amsterdam

From these hotels, the venue can be reached in 15-30 minutes with public transport. In all public transportation, you can check in and out with a Mastercard or Visa contactless credit card and also with Apple Pay and Google Wallet.

Venue

The conference will be held in the Turing room at the Congress Centre of Amsterdam Science Park, next to Centrum Wiskunde & Informatica (CWI).

Address: Science Park 125, 1098 XG Amsterdam

See here for location in Google Maps.

Organizers

• Karen Aardal
• Antonios Antoniadis
• Daniel Dadush
• Dick den Hertog
• Ruben Hoeksma
• Leo van Iersel
• Etienne de Klerk
• Monique Laurent
• Guido Schäfer
• Leen Stougie
• Marc Uetz
• Bert Zwart

CWI Research Semester Programme

This PhD School is part of the broader CWI Research Semester Programme on Learning Enhanced Optimization, contributing to its overarching mission of advancing cutting-edge research in theoretical computer science, operations research and beyond.

See here for more information about the whole research semester programme

Financial Support

We gratefully acknowledge the financial support of CWI whose contributions helped to make this event possible.