CS 325: Topics in Computational Sustainability, Spring 2016

Instructor: Stefano Ermon

Course Description

Computational Sustainability focuses on developing computational models, methods and tools to support policy makers in developing more effective policies for sustainable development. In this course, we will study recent computational approaches that have contributed to addressing sustainability topics related to biodiversity, climate, environment, urban design, transportation, buildings and others. Computational themes include machine learning, optimization, statistical modeling, and data mining.

News

Date
04.4.16 Assignments deadlines updated.
03.25.16 Piazza online
03.25.16 Webpage online

General Information

Lectures: Tuesdays, Thursdays 10:30-11:50
Office hours: Thursdays, 1-2 pm, Gates 228 or by appointment
Piazza: Please sign up on Piazza

(Tentative) Syllabus

Lecture Topics Lecture Slides Readings etc.
Introduction [slides] [video] [paper]
Linear Regression [slides] [Chapter 3, Pages 44-45]
ML for Poverty mapping [slides] []
Nonlinear regression [slides] []
Linear Programming [slides] []
Integer Programming [slides] []
Landscape conservation [slides] []
Guest Lecture (4/26): Sustainable data centers [slides] []
Guest Lecture (4/28): Smart Buildings [slides] []
Guest Lecture (5/3): Nature Capital project [] [Website]
Stochastic Optimization [Slides]
Markov Decision Processes [Slides]
Guest Lecture: Big Data and food security [] []

Many thanks to Carla Gomes, Zico Kolter and Daniel Sheldon for sharing teaching materials.

Coursework

The main graded components are:

  • A reaction paper critically summarizing a sustainability-related problem and published solution approaches (20%). You can choose to use the reaction paper as background research for your project. Some ideas are provided below.
    • 1 or 2 pages max, due in class April 28.
    • Critically summarize the reading material. To develop your argument, you can:
      • compare the work to alternative approaches
      • discuss how could the work be improved
      • discuss how could the work be expanded or applied to other domains
      • find strengths and weaknesses
  • A course project (60%). You can work individually or in a small team (3 people max).
    • Project proposal (max 1 page): Due May 3 (in class)
    • Final project report (max 5 pages): Due May 31 June 5
  • An in-class presentation (20%). Please refer to the email you received about schedule and details.

Bibliography

Here you can find some references for papers that are relevant for the reaction paper assignment. You are welcome to choose a paper that is not in this list.

Conference Proceedings:

AAAI 2016: Computational Sustainability and AI (scroll down to find list of accepted papers here)

AAAI 2015: Computational Sustainability and AI (scroll down to find list of accepted papers here)

AAAI 2014: Computational Sustainability and AI (scroll down to find list of accepted papers here)

AAAI 2013: Special Track on Computational Sustainability and AI (scroll down to find list of accepted papers here)

IJCAI 2013: Special Track on AI and Computational Sustainability (scroll down to find list of accepted papers here)

AAAI 2012: Special Track on Computational Sustainability and AI (scroll down to find list of accepted papers here)

AAAI 2011: Special Track on Computational Sustainability and AI (scroll down to find list of accepted papers here)

International Conference on Computational Sustainability CompSust’09 (link)

CompSust’10 Papers (link)

CompSust’12 Papers (link)

NIPS 2013 Workshop: Machine Learning for Sustainability (link)

KDD 2012 Workshop on Data Mining Applications in Sustainability (link)

KDD 2011 Workshop on Data Mining Applications in Sustainability (link)

Other bibliographies:

CS 691(link)

CSE 8803(link)

Project Ideas and datasets

[Dataset| Climate and Food Security]

How well can we predict crop yields as a function of weather and climate? The first link below provides data generated by a detailed crop simulator (courtesy of David Lobell) that you can use as a starting point. Successful models can then be tested on real world data from satellites. Accurate models could be used to improve our understanding of poverty and hunger in developing countries as well as to better inform farming practices (precision agricolture). The second link proposes an alternative approach for predicting aggregate yields at the county or national level using different data sources.
Link
Link

[Dataset| Poverty mitigation]

Eradicating worldwide poverty by 2030 is the top goal on the United Nations’ sustainable development agenda, published late last year. But a lack of data has frustrated efforts to measure progress toward the goal. Is it possible to accurately predict economic indicators such expenditures and asset-based measures of wealth from high resolution satellite images? Data set provided by Neal Jean.
Link

[Dataset|Challenge in Conservation]

Large Landscape Conservation - Synthetic and Real-World Datasets Bistra Dilkina, Katherine Lai, Ronan Le Bras, Yexiang Xue, Carla P. Gomes, Ashish Sabharwal, Jordan Suter, Kevin S. McKelvey, Michael K. Schwartz and Claire Montgomery. AAAI-13: AAAI Conference on Artificial Intelligence
Link 1
Link 2

[Dataset | Challenge in Conservation | POMDP]

Adaptive management of migratory birds under sea level rise. Nicol S, Iwamura T, Buffet O, Chadès I. International Joint Conference on Artificial Intelligence (IJCAI)
Link

[Dataset | Challenge in Climate]

Forecast Oriented Classification of Spatio-Temporal Extreme Events. Z. Chen, Y. Xie, Y. Cheng, K. Zhang, A. Agrawal, W. Liao, N. F. Samatova, and A. Choudhary. International Joint Conference on Artificial Intelligence (IJCAI)
Link

[Home Energy]

REDD: A public data set for energy disaggregation research. Zico Kolter and Matthew J. Johnson. SustKDD: Workshop on Data Mining Applications in Sustainability
Link

[Home Energy]

Smart*: An Open Data Set and Tools for Enabling Research in Sustainable Homes. Sean Barker, Aditya Mishra, David Irwin, Emmanuel Cecchet, Prashant Shenoy, and Jeannie Albrecht. SustKDD: Workshop on Data Mining Applications in Sustainability, 2012.
Link

[Smart cities | Bike Share]

Balancing bike sharing systems (BBSS): instance generation from the CitiBike NYC data. arXiv preprint arXiv:1312.3971 Urli, Tommaso.
Link

Other Resources

CompSustNet

Videos

Similar classes taught at other universities