Home

2023 YinzOR Student Conference

The CMU INFORMS Student Chapter is excited to host the 6th YinzOR Conference. This student conference provides an opportunity for students whose research is in Operations Research, Management Science, Industrial Engineering, and other related fields to interact with each other. This year, we are hosting the conference In-Person at the Tepper School of Business at CMU on August 25th-26th!

Information

Registration is FREE and now open for the In-Person YinzOR 2023 conference!
Register by August 15 to attend.

Flash Talk and Poster Competition deadline (date to come). For accepted flash talks and posters, CMU INFORMS chapter will provide accommodations if needed to help with travel and printing costs. More information on accomodations to come. Please click on the individual poster and flash talk tabs for additional information and guidelines.

Poster Competition Winners!

1rd place - $500 - Wenbin Zhou - Counterfactual Generative Models for Time-Varying Treatment
2rd place - $200 - Mo Liu - Active Learning in the Predict-then-Optimize Framework: A Margin-Based Approach
3rd place - $100 - Johnathan Vicente - Computing Equilibrium Automotive Technology Decisions Under Regularization
Fan Favorite - $100 - Hatice Gökçen Güner - Machine Learning Driven Shape Optimization of Metamaterials with Adaptable Force-Displacement Characteristics for Soft Robots

Flash Talk Competition Winners!

Top 3 Presentations - $100 - (No Order)

Vikas Deep - Optimal Adaptive Experimental Design of Inference of Average Treatment Effect
Miaolan Xie - Reliable Adaptive Stochastic Optimization for Messy Data with High Probability Guarantees
Amine Bennouna - Holistic Robust Machine Learning and Data-Driven Decision-Making


Fan Favorite - $100

Mo Liu - Pricing under the Generalized Markov Chain Choice Model: Learning through Large-scale Click Behaviors

Organizing Committee

CMU OR/OM/ChemE PhD Students

Nilsu Uzunlar (Co-Chair) (email: nuzunlar@andrew.cmu.edu )
Tian Wang (Co-Chair) (email: tianw2@andrew.cmu.edu )
Tom Krumpolc (Webmaster Co-Chair)
Alex Lim (Webmaster Co-Chair)
Lin An (Speaker Co-Chair)
Mik Zlatin (Speaker Co-Chair)
H. Satyam Verma (Marketing Chair)
Siyue Liu (Flash Talk and Poster Session Co-Chair)
Vrishabh Patil (Flash Talk and Poster Session Co-Chair)
Macarena Navarro (Logistics Co-Chair)
Sebastian Vasquez (Logistics Co-Chair)
Anthony Karahalios (Sponsor Chair)

* Please feel free to email either co-chair with any of your questions/comments!

Past Conferences

YinzOR 2022
YinzOR 2021
YinzOR 2019
YinzOR 2018
YinzOR 2017

Conference Program

Friday, August 25th

1:00pm-1:30pm Registration
Come pick up your name tag before the conference begins! All talks will take place in Tepper Room 3801.
1:30pm-1:45pm Opening Remarks
Willem-Jan van Hoeve, Carnegie Bosch Professor of Operations Research and Senior Associate Dean, Tepper CMU
1:45pm-2:30pm LocalSolver
Léa Blaise, Optimization scientist at LocalSolver
The presentation will give a global introduction to LocalSolver, giving the basics of LocalSolver's modeling features as well as an overview of the various solving techniques implemented inside the solver, with an emphasis on scheduling problems. The presentation will conclude with an example of a real-life optimization problem and how LocalSolver solves it.
2:30pm-3:15pm Machine Learning for Global Optimization
Can Li, Assistant Professor of Chemical Engineering, Purdue University
Global optimization is a well-established area in operations research and computer science. Until recently, its methods have focused on solving problem instances in isolation, ignoring that they often stem from related data distributions in practice. However, recent years have seen a surge of interest in using machine learning as a new approach for solving optimization problems, either directly as solvers or by enhancing exact solvers. In this talk, we will give an overview of recent approaches for accelerating optimization algorithms using machine learning. Examples will include learning to branch with graph convolutional neural networks, learning to select convex relaxations in global optimization algorithms, and learning to solve AC optimal power problems using a difference of convex relaxations.
3:15pm-4:20pm (2nd Floor) Poster Session
Students will setup posters in an open space. Winners will be determined by both formal judges and attendees popular votes!
4:30pm-5:15pm Velocity-aware inventory placement
Cristiana Lara, Senior Research Scientist at Amazon
Amazon offers millions of unique items for sale on its website. The popularity of these items is not evenly distributed, and the mean weekly demand for different products on Amazon spans multiple orders of magnitude. Because of this enormous scale and selection, deciding how to optimally place this inventory throughout our fulfillment network is very challenging. Directly modeling all products is not feasible without advanced decomposition techniques and massive compute power. On the other hand, down-sampling products typically does not adequately capture the behavior of the “long tail”. We propose a scalable method for modeling both head and tail demand, and placing the inventory to best utilize our network assets. Our model matches the ship and storage capacity of nodes in our network with the velocity profile of demand in order to minimize fulfillment cost. We use this framework to derive insights on the relationship between placement strategy and customer experience.
5:15pm-5:45pm Coffee Break and Small Group Chats
We will break attendees up into small groups so that people can get a chance to know each other a little better!
5:45pm-6:30pm Project 412: Bridging Students to Communities
Michael Hamilton, Assistant Professor of Business Analytics and Operations at Katz Graduate School of Business, University of Pittsburgh
We investigate some of the challenges faced by Black-owned businesses, focusing on the unique geographic inequities present in Pittsburgh. We determine that university students represent an under-utilized market for these businesses. We investigate the root causes for this inefficiency and design and deploy a platform, 412Connect, to increase support for Pittsburgh Black-owned businesses from within the university community. The platform operates by coordinating interactions between student users and participating businesses. Our platform design choices are aided by two simple, novel models: a model for badge design and a model for equity-orientated recommendation. We analyze and optimize the badge structure and propose a simple, robust, and equitable recommendation policy. Finally, we present preliminary results on the platform's impact during its runs in Fall 2021 and Fall 2022.
Coauthors: Alex DiChristofano (WUSTL), Qiqi Hao (University of Pittsburgh), Sera Linardi (University of Pittsburgh)
6:30pm-9:00pm Dinner
All are welcome to a free dinner catered by a local Pittsburgh restaurant. We may go out for drinks after also.

Saturday, August 26th

9:00am Registration Opens
Come pick up your name tag before the second day of the conference begins!
9:45am-10:15am Breakfast
All are welcome to a free breakfast and coffee to start the day
10:15am-11:00am Intertemporal Pricing in the Presence of Consumer Behaviors
Hansheng Jiang, Incoming Assistant Professor at Rotman School of Management, University of Toronto
Dynamic pricing policies are widely implemented in modern retailing platforms, where consumer behaviors significantly influence the outcomes of these policies. The complexity and non-stationarity introduced by consumer strategic or boundedly rational behaviors call for more tailored pricing analytic methodologies in both demand learning and pricing optimization. Rooted in the prospect theory, consumer reference effects suggest that consumers' decisions may depend on not just current prices but also historical or external prices. In my talk, I will discuss our work on intertemporal pricing under heterogeneous reference effects. We spotlight the sub-optimal nature of constant pricing and illustrate how considering consumer heterogeneity can improve retailer revenue. Our empirical validation with JD.com data reveals the risk of revenue loss when overlooking consumer heterogeneity and underscores the importance of promotions and price fluctuations due to these heterogeneous reference effects. Time permitting, I will extend this discussion to a multi-product context and provide characterizations of optimal and myopic pricing policies.
11:00am-11:15am Coffee Break
11:15am-12:00pm Mixed-Integer Optimization with Constraint Learning
Holly Wiberg, Incoming Assistant Professor of Public Policy and Operations Research at Heinz College, Carnegie Mellon University
In this talk, we introduce a framework for optimization with learned constraints (OptiCL) that embeds trained machine learning models directly into mixed-integer optimization formulations. We train machine learning models to approximate functional relationships between decisions and outcomes of interest and subsequently optimize decisions under these data-driven learned constraints and/or objectives. We also introduce two approaches for handling the inherent uncertainty of learning from data. First, we characterize a decision trust region using the convex hull of the observations, to ensure credible recommendations and avoid extrapolation. Then, we propose an ensemble learning approach that enforces constraint satisfaction over multiple bootstrapped estimators or multiple algorithms. The OptiCL framework is particularly applicable in healthcare, from prescribing clinical treatments to designing public health policies, which often involve complex decisions with no deterministic relationship to the outcomes of interest. We present two case studies: (1) chemotherapy regimen optimization, in which we formulate a mixed-integer programming model to identify promising treatment combinations using ML-derived toxicity constraints, and (2) food basket composition for humanitarian aid under a palatability constraint.
Co-authors: Donato Maragno, Dimitris Bertsimas, Ilker Birbil, Dick den Hertog, and Ade Fajemisin
12:00pm-1:00pm Lunch
Lunch on your own! We welcome you to explore restaurants around the area. Please ask us if you would like any recommendations!
1:00pm-2:30pm 10-Minute Flash Talk Competition
Selected students will present 10-Minute Flash Talks in a competition judged by the attendees with cash prizes!
2:30pm-2:45pm Coffee Break
2:45pm-3:30pm Balanced Filtering via Non-Disclosive Proxies
Emily Diana, Incoming Research Assistant Professor at TTIC in 2023, Incoming Assistant Professor at Tepper School of Business in 2024
We study the problem of non-disclosively collecting a sample of data that is balanced with respect to sensitive groups when group membership is unavailable or prohibited from use at collection time. Specifically, our collection mechanism does not reveal significantly more about group membership of any individual sample than can be ascertained from base rates alone. To do this, we adopt a fairness pipeline perspective, in which a learner can use a small set of labeled data to train a proxy function that can later be used for this filtering task. We then associate the range of the proxy function with sampling probabilities; given a new candidate, we classify it using our proxy function, and then select it for our sample with probability proportional to the sampling probability corresponding to its proxy classification. Importantly, we require that the proxy classification itself not reveal significant information about the sensitive group membership of any individual sample (i.e., it should be sufficiently non-disclosive). We show that under modest algorithmic assumptions, we find such a proxy in a sample- and oracle-efficient manner. Finally, we experimentally evaluate our algorithm and analyze generalization properties.
3:30pm-3:45pm Short Break
3:45pm-4:30pm Provably Optimal Reinforcement Learning for Inventory Problems with Unknown Cyclic Demands
Evelyn Gong, Incoming Assistant Professor at Tepper School of Business, Carnegie Mellon University
Motivated by a long-standing gap in inventory theory, we invent reinforcement learning methods with provable optimality guarantees for inventory management problems with unknown cyclic demand distributions. We design provably efficient algorithms that leverage the structure of inventory problems. We apply the standard performance measure in online learning literature, regret, i.e. the difference between the total expected cost of our policy and the total expected cost of the clairvoyant optimal policy that has full knowledge of the demand distributions a priori. Our policies achieve optimal regret for a number of models. We remove the regret dependence on the cardinality of the state-action space, which is an improvement over existing RL algorithms (if applied on our inventory problems). We conducted experiments with a real sales dataset from Rossmann. Our policy converges rapidly to the optimal policy and dramatically outperforms the best policy that models demand as i.i.d. instead of cyclic.
4:30pm-4:45pm Closing Remarks
Fatma Kilinc-Karzan, Associate Professor of Operations Research and CMU INFORMS Faculty Advisor
5:00pm-7:30pm Happy Hour
Free refreshments to celebrate a wonderful conference will be available in TEP2001, TEP2002, and TEP2003. The party may continue past 7:30pm outside of the building.

10-Minute Flash Talk Competition

Holistic Robust Machine Learning and Data-Driven Decision-Making
Amine Bennouna, Massachusetts Institute of Technology
Proactive capacity planning of transmission and redispatch of generation systems to prevent electricity supply disruptions
Seolhee Cho, Carnegie Mellon University
Pricing Strategies for Online Dating Platforms
Titing Cui, University of Pittsburgh
Optimal Adaptive Experimental Design of Inference of Average Treatment Effect
Vikas Deep, Northwestern University
A 4/3-Approximation Algorithm for Half-Integral Cycle Cut Instances of the TSP
Billy Jin, Cornell University
Nonprogressive Diffusion on Social Networks: Approximation and Applications
Yunduan Lin, University of California, Berkeley
Pricing under the Generalized Markov Chain Choice Model: Learning through Large-scale Click Behaviors
Mo Liu, University of California, Berkeley
High Probability Complexity Bounds for Adaptive Optimization Methods with Stochastic Oracles
Miaolan Xie, Cornell University
Estimate-Then-Optimize versus Integrated-Estimation-Optimization versus Sample Average Approximation: A Stochastic Dominance Perspective
Haofeng Zhang, Columbia University

Saturday, August 26th 1:00pm-2:30pm

Register here for the 10-minute flash talk competition!


Prizes, Rules, Judges, and Logistics

Prizes: Top 3 places will receive $100 each. These are determined by the judges. One fan favorite will win $100, voted on by the students.

Rules: Each participant will have 7 minutes to present their work. We will have 3 minutes for questions after each presentation. Each talk will receive scores from the judges. You may present using slides or the whiteboard.

Judges: Michael Hamilton, Sridhar R. Tayur, R. Ravi. The fan favorite will be voted on by students.

DEADLINE: August 9

Logistics: Provide powerpoint slides at least 1 week before the event. The slides can be emailed to vmpatil (atsign) andrew.cmu.edu

Accommodations can be provided for one night (August 25) if needed for accepted flash talk presenters. Hotel rooms will be shared with one other person. We cannot provide travel reimbursements.

Poster Competition

Corrected wind speed prediction via multi-resolution kriging
Zheng Dong, Georgia Institute of Technology
Repair or Replace: Evidence of Sequential Diagnostic Decisions on Product Returns
Jingxuan Geng, Temple University
Machine Learning Driven Shape Optimization of Metamaterials with Adaptable Force-Displacement Characteristics for Soft Robots
Hatice Gökçen Güner, Carnegie Mellon University
Strategic Delay in Grocery Delivery Platforms
Hao Jiang, Temple University
Compartmental Modeling for Power Outage Prediction using Neural Stochastic Differential Equations
Xiangrui (Cindy) Kong, Carnegie Mellon University
Content Promotion for Online Content Platforms with the Diffusion Effect
Yunduan Lin, University of California, Berkeley
Active Learning in the Predict-then-Optimize Framework: A Margin-Based Approach
Mo Liu, University of California, Berkeley
Customized or Standardized? Design Optimization for Smart Mobile Lockers in E-commerce Last Mile Deliveries
Si Liu, McMaster University
Fast food stores with a drive-through option recovered from COVID-19, stores without did not
Partha Sarathi Mishra, Northwestern University
Decision Diagrams in Space!
Isaac Rudich, Montréal Polytechnique
Computing Equilibrium Automotive Technology Decisions Under Regulation
Johnathan Vicente, Carnegie Mellon University
Entropic Regularization for Adversarial Robust Learning
Jie Wang, Georgia Institute of Technology
Counterfactual Generative Models for Time-Varying Treatments
Wenbin Zhou, Carnegie Mellon University

Friday, August 25th 3:00pm-4:30pm

Register here for the poster competition!


Prizes, Rules, Judges, and Logistics

Prizes:

Rules: Accepted students will present a standard sized poster. They will have ~5 minutes to present their poster to the judges.

Judges: Cristiana Lara, Alan Scheller-Wolf, Gérard Cornuéjols. A fan favorite will be voted on by students.

DEADLINE: August 9

Logistics: Please provide posters at least 1 week before the event. The posters will be printed by the CMU INFORMS chapter and late corrections will not be accepted. The posters can be emailed to siyueliu (atsign) andrew.cmu.edu

Accommodations can be provided for one night (August 25) if needed for accepted poster presenters. Hotel rooms will be shared with one other person. We cannot provide travel reimbursements.

© CMU INFORMS