Inaugural Data Impact Summer Institute kicked off today with 60+ students 15 projects & 19 faculty project mentors.The Data For Impact Summer Institute, planned in collaboration with the Martindale Center and I-DISC, is an eight week program that will run from June 15th - August 7th.
I-DISC Faculty Members (in bold) are mentors and supporting students in the following Data for Impact Projects:
Data to Control: Towards Data-Driven Model Predictive Control for Chemical Process Automation
Most chemical and biological processes are dynamical systems. This means that their state variables (i.e. variables that characterize what state the system is in) are continuously changing, often underlined by highly nonlinear correlated behavior that many not be easily captured by physics-based models. Modern plants in the energy and chemical industry have advanced data acquisition technologies, enabled in many cases by solutions offered by OSISoft LLC, the industrial partners on this project. These technologies allow for collecting, storing, and analyzing data from thousands of sensors every second (or faster). Our ultimate goal is to leverage this data to design, optimize, and control new energy and chemical systems. We will begin addressing this larger goal by developing algorithms that will allow us to extract the underlying ordinary differential equations from time-varying data. This algorithm will then allow us to take time-varying plant data and build data-driven dynamic equations that accurately captures the overall process. We specifically intend to build on the state-of-the-art algorithms from the applied mathematics community on inferring equations from data that have been successfully applied in the fluid mechanics domain by incorporating a number of new features including the concept of infusing chemical engineering domain knowledge as constraints while training the data-driven model.
Mayuresh Kothare, Professor & Chair, Chemical & Biomolecular Engineering
Srinivas Rangarajan, Assistant Professor, Chemical & Biomolecular Engineering
Aziz Alsalem '22, Chemical Engineering
Nick Kosir '20, IDEAS
Jingming Shi '23, Chemical Engineering
From Molecules to Medicine: Overcoming the Time Scale Challenge
Biological processes underpinning human wellness occur over seconds, hours, days, and longer, yet the governing molecular mechanisms occur on time scales from picoseconds to microseconds. Molecular mechanistic understanding of complex biological systems can dramatically impact disease diagnosis and treatment. Even the longest simulations that resolve matter at the atomic scale can only examine atomistic behavior over micro- or perhaps milliseconds. Advanced data processing techniques have emerged that hold promise for the ability to bridge information obtained from molecular-scale descriptions of matter to address questions that manifest at human physiological time scales. In this project, students will develop an understanding of structure/function relationships in biology and the intrinsic multi-time scale nature of addressing human wellness from a molecular point of view. The team will learn specific chemical-physical structure-function coupling mechanisms in the human blood protein von Willebrand Factor (vWF), which is potentially implicated in bleeding disorders affecting ~2% of the human population. Team members will use molecular scale computational simulations, in conjunction with advanced data processing techniques, to understand how data methods are being used to bridge molecular scale mechanistic information and impact treatment of conditions at the human physiological scale.
Ed Webb, Associate Professor, Mechanical Engineering & Mechanics
Graduate Project Assistant: Sagar Kania, Mechanical Engineering
Hassan Al Kawalden '21, Mechanical Engineering
Andrew Donnachie '21, Statistics
Ching Laam (Damon) Luk '21, Electrical Engineering
Spam Spotting: Using AI Tools to Educate and Improve Online Decision-Making
On websites like Amazon and TripAdvisor, fake reviews (“spams”) are prevalent. Stories about spams and their victims have been reported widely; these spams overturn product and service reputations and adversely affect users’ decision making. To protect the general public, AI-based spam detectors have been employed to actively flag the spams. Also, more sophisticated users may use their judgments to spot spams. However, AI detectors are not always accurate and transparent, and will not be much trusted and adopted by the general public for fighting spams (“algorithm aversion”). Further, without training, even sophisticated users have difficulty in distinguishing spams from genuine reviews. This project will work towards an education-based defense against spams, where the general public will be educated to acquire the skills to spot spams, and to trust and properly rely on AI detectors to improve their protection. Our summer scope of work will be 1) develop surveys and questionnaires to understand the scope of the challenges; 2) code a role-playing game where a spammer can craft spams for the spotters to catch, both for fun and for research; 3) code a simple tutoring tool to teach human to use AI spam detectors.
Sihong Xie, Assistant Professor, Computer Science & Engineering
Qiong Fu, Professor of Practice, College of Education
Graduate Project Assistant:
Rui Chen, Special Education
Jack Curtis '22, Psychology / Computer Science
Diana Garcia '21, Psychology
Jennifer Liu '23, Computer Science
Griffin Reichert '21, Computer Science / Economics
Yifan Zhang '22, Computer Science
Real-Time Machine Learning in Experimental Materials Science
In materials science and physics more broadly there is a growing trend to conduct multimodal experiments (experiments that collect data from a variety of sources). The boon in data collection has left a majority of the data collected under-analyzed leaving important physics left undiscovered. This project will develop machine and deep learning methods to discover actionable information from such data. This project will also consider how such models can be implemented on specialty AI hardware for real-time analysis. The work will focus on materials problems as they provide unique ways to stress-test practical theories of machine and deep learning. Outcomes of this work have direct impacts on creating interpretable AI, controlling fairness and bias, and creating autonomous control systems. The impacts of these theories can be adapted to solve problems in medicine and healthcare, resource management and logistics, and manufacturing and processing.
Joshua Agar, Assistant Professor, Materials Science & Engineering
Tri Nguyen '21, Materials Science & Engineering
Oluwafolajinmi Olugbodi '23, Computer Engineering
William Reichard-Flynn '21, Earth & Environmental Science
Andrew Zheng '24, Mechanical Engineering
A Dose of AI for Disease Prevention and Treatment
AI combined with predictive analysis has helped change the landscape of disease prevention and treatment – bringing a paradigm shift to healthcare. With improved image analytics, concrete clinical and diagnostic decision-making, AI has been highly beneficial for the treatment of chronic diseases like cancer, neurology, and cardiology. In this project, students will develop predictive AI algorithms for early detection and biomarker discovery on various diseases over different kinds of data, such as mood disorder prediction in mobile keyboard, ROP and Parkinson's disease prediction with image and clinical data. Students will engage in focused problems to develop prototypes, run experiments and scale how this works, etc. This project would invigorate fields like computer science and healthcare and trigger the development of new models and algorithms. It could also help solve some medical problems for the benefit of humanity as well as harness the power AI in healthcare. Through this process, students will begin to form lasting AI capability in healthcare.
Lifang He, Assistant Professor, Computer Science & Engineering
Dinglun He '20, Electrical & Computer Engineering
Kenny Lin '22, Computer Science & Engineering
JiaBei Wei '22, Computer Science & Engineering
Danyang Zhu '21, Statistics
The “Falling Knife” Project
The goal of this project is to build a software with customer interface to detect whether or not the stock that a user has selected is a “falling knife.” If the stock is a falling knife, how far is the fall, and when will it go back up to its previous price level? Students will understand what a falling knife is, how to find these falling knives, and to teach machine learning the pattern of them. Lastly, students will build the base model on findings and back test its accuracy. Currently the project uses Technical Analysis to identify those short term (days, weeks) falling knives with MACD, RSI, EMA, and the fall duration. The team will cross check the S&P 500 for the past 30 years to see if findings hold, and use fundamental analysis and financial matrices to identify what counts as a long term (Months, years) falling knife. To see more about the project, see a 30-minute video from Masters in Financial Engineering students, here.
Patrick Zoro, Professor of Practice, Finance
Ethan Leifer '24, undecided
Shiqi Liu '21, Statistics
Eric Meskin '21, Finance
Odilon Niyomugabo '21, Industrial & Systems Engineering
Delton Tschuda '21, Finance / Industrial & Systems Engineering
Full details of all projects, faculty & students can be found on the Creative Inquiry Website
Stay tuned for announcements about Press Conferences and other presentations throughout the summer. Questions about the institute can be sent to firstname.lastname@example.org.