Goals

Structural modeling and machine learning methods are being increasingly used in research on technology. The purpose of the workshop is to introduce Ph.D. students/researchers to the integrated framework of structural modeling and machine learning. As most of the researchers involved in these methodological domains would agree, the upfront cost of learning and applying structural modeling and machine learning is a quite steep. The workshop would aim to reduce the upfront cost of learning by providing several hands-on tutorials on structural modeling and machine learning. The objective of the workshop is also to open up common ground for future discussions between Ph.D. students and colleagues across schools who use/ wish to use structural modeling and machine learning methods in information systems research.

WORKSHOP FORMAT

Several hands-on tutorials (directed primarily at doctoral students) on structural modeling and machine learning.

  • Structural Demand Estimation Using Aggregate Data
  • Single Agent Dynamic Structural Optimization Model
  • Dynamic Structural Game Model
  • Bayesian Learning Model
  • Structural Estimation of Auctions
  • Matching Models
  • Hidden Markov Models
  • Word Embeddings (Word2Vec and Glove)
  • Variational Probabilistic Inference
  • Decision-making under Uncertainty and Personalization

WORKSHOP CHAIRS

Anindya Ghose - Stern School, NYU

Param Vir Singh - Tepper School, CMU

Yong Tan - Foster School, UW

2017
August 5-6

University of Washington
Seattle

40
Students

8
Instructors

PH.D. STUDENT SELECTION

We will admit a total of 40 students only. Each student applicant will have to submit a short email from his/her advisor supporting his/her case for attending the SMART workshop. Letters could be sent to Yong Tan at ytan@uw.edu

The requirements that a PhD student must satisfy:

  • must be in second year or later at the time of attending the workshop;
  • should have taken PhD level Econometrics courses which includes working with discrete choice models and coding the estimation procedure (rather than using Stata or any other pre-packaged econometrics software);
  • should have taken a PhD level game theory course;
  • should have taken a course that involves stochastic modeling (should be proficient with probability calculations);
  • should be proficient in programming language (preferably Matlab for Structural Modeling, and Python for machine learning).

If there are more than two students who apply from the same school, we will ask the school to rank the students. Due to limited slots, we guarantee top 2 will be included, but depending on demand others may or may not be included in the workshop. A student who has already taken a structural course (and gets a recommendation from the instructor) or is using structural methods in his/her research will be given priority.

Instructors

Panagiotis (Panos) Adamopoulos

Emory University

VILMA TODRI

Emory University

Yan Huang

University of Michigan

Julian Guo

Michigan State University

Yingfei Wang

University of Washington

Stephanie Lee

University of Washington

Hema Yoganarasimhan

University of Washington

Junming Yin

University of Arizona

Program

Day 1: Saturday, August 5, 2017

8:30 - 9:00 a.m.Continental Breakfast
9:00 - 9:15 a.m. Welcome Remarks (All sessions are held in Paccar Hall 290 unless specified otherwise.)
9:15 – 11:45 a.m. Dynamic Structural Models (Hema Yoganarasimhan)
noon - 1:00 p.m. Boxes Lunches in Anthony’s Forum (DEM 302)
1:00 – 3:30 p.m. Structural Demand Estimation Using Aggregate Data and Bayesian Learning Model (Yan Huang)
3:30 - 3:45 p.m. Coffee Break
3:45 - 5:00 p.m. Structural Estimation of Auctions (Julian Guo)
5:00 p.m. (2) shuttle buses at Steven’s Way will take us to South Lake Union - there will be signage and volunteers showing the way
5:45 - 9:00 p.m. Dinner and Reception from Argosy Cruises
9:00 p.m. (2) shuttle buses return to Paccar Hall

Day 2: Sunday, August 6, 2017

8:30 - 9:00 a.m. Continental Breakfast
9:00 - 10:15 a.m Matching Models (Stephanie Lee)
10:15 - 10:30 a.m.  Coffee break
10:30 - 11: 45 a.m. Hidden Markov Models (Vilma Todri)
Noon  - 1:00 p.m. Boxed Lunches in Anthony’s Forum (DEM 302)
1:30 – 2:45 p.m. Variational probabilistic inference (Junming Yin)
2:45 - 3:00 p.m. Coffee break
3:00 - 4:15 p.m. Decision-making under uncertainty and personalization (Yingfei Wang)
4:15 - 5:30 p.m. Word Embedding (Panos)
5:30 - 5: 45 p.m. Concluding Remarks/ Workshop Ends

Software and Packages

As you prepare for the attending the workshop, please note that the tutorials would be hands-on and would involve lecturing and coding. Please remember to bring your Laptops to the tutorials. The coding will be done in Matlab and Python.

For Matlab, if you do not already have Matlab Econometrics toolbox, you can download it for free from http://www.spatial-econometrics.com/.

For Python, please install Python 2.7 along with the following packages:

Any standard Python distribution, such as Anaconda, should contain most of these packages, if not all. If you need help installing these packages, please refer to the documentation of the distribution you installed or these generic instructions https://programminghistorian.org/lessons/installing-python-modules-pip

Social Events

Lake Washington Cruise

During this 1½-hour narrated cruise of Lake Washington, you’ll learn the history of the area and cruise past magnificent lakeside properties as you relax and enjoy one of the world’s truly beautiful urban lakes.

Your experience includes:
  • Panoramic views of onshore properties plus the natural beauty of the lake and shore.
  • History of the Eastside and Seattle neighborhoods bordering Lake Washington, take a swing by the Bill and Melinda Gates estate and pass under the new Evergreen Point Floating Bridge (SR 520).
  • Full-service bar available on board for the purchase of local wine and beer, non-alcoholic beverages, and snacks.

Venue

Paccar Hall
Foster School
4273 E Stevens Way NE
Seattle, WA 98915