M.S. Thesis: Changing Edges in Graphical Model Algorithms
Stochastic Information Processing Systems (with SONIC)
This research area seeks to establish statistical foundations for information processing to achieve energy efficiency and robustness in emerging nano-scale beyond-CMOS functional fabrics, which are physically stochastic. It will do so by matching the fabrics’ inherent statistical attributes to those of emerging machine learning frameworks for statistical inference (which are functionally stochastic). The goals of this research area are to achieve orders-of-magnitude improvements in energy-efficiency and robustness for compute-intensive, sensing, and communication-intensive tasks implemented in beyond-CMOS technologies.
Publications:
- L. Chang, A. Chatterjee, and L. R. Varshney, “Performance of LDPC Decoders with Missing Connections,” IEEE Transactions on Communications, vol. 65, no. 2, pp. 511-524, February 2017. [pdf]
- L. Chang, A. Chatterjee, and L. R. Varshney, “LDPC Decoders with Missing Connections,” in Proceedings of the 2016 IEEE International Symposium on Information Theory (ISIT), Barcelona, Spain, 10-15 July 2016. [pdf]
Data and Networks
– Computational Gastronomy on Graphs
This project seeks to use data on food chemistry, including flavor, texture, and olfactory, to understand the scientific side of culinary practice and human sensory perception of food. Motivations of this research project includes creating novel recipes to encourage people to eat in a healthy way to reduce obesity [1],[2], and improving the experience for people who have dietary restrictions or are constrained by environment limitations such as hospital patience, astronauts, and those with allergies. Interested researcher are encouraged to learn about the Food and Data Workshop hosted by my advisor and our committee!
Presentation:
- L. Chang and L. R. Varshney, “World Culture of Food Texture Networks,” Food and Data Workshop, 2016
Interesting Related Projects:
- IBM Chef Watson
- Appetit by Telefonica
- Project Nourished
- Olfactory Virtual Reality Headset
- Dietary-Aware Dining Table
– Sports Analytics with Dynamic Network Models
This project uses tracking data of sporting events and combines dynamic network analysis and control theory to perform various kinds of analysis that is not obvious or easily to be characterized by scores of the matches. Motivations of the research project includes providing insights to team coaches, final score prediction, and real-time strategy recommendations, etc.
Poster:
- L. Chang, “Use ‘Redundancy’ in Graphs for Soccer Match Analysis: Understand the Collective Intelligence among Agents,” IBM Cognitive Colloquium, 2016
Interesting Related Misc:
- Predicting Football Matches Using Data With Jordan Tigani – Strata Europe 2014 Talk by Google Engineer
- Tuning in on Noise? Blogpost by Prof. Tom Murphy at the University of California, San Diego
Reference:
[1] F. Pinel and L. R. Varshney, “Computational Creativity for Culinary Recipes,” in Proceedings of the ACM CHI Conference on Human Factors in Computing Systems, Toronto, Canada, 26 April – 1 May 2014. (Link)
[2] H. Bai, R. Chunara, and L. R. Varshney, “Social Capital Deserts: Obesity Surveillance using a Location-Based Social Network,” in Proceedings of the Data for Good Exchange (D4GX), New York, New York, 28 September 2015. (NYC Media Lab – Bloomberg Data for Good Exchange Paper Award) (Link)