Roux Summer Research Internship - Causal Machine Learning in Health and Medicine Introduction
About the Opportunity
Summer Research Internships at The Roux Institute
The Roux Institute Summer Research Internship Program is a 10-week full-time internship for graduate and undergraduate learners on-site at the Roux Institute campus of Northeastern University in Portland, ME.
- Stipend: $5,000 for Undergraduate Research Interns $7,500 for Masters & PhD Research Interns.
- Gain interdisciplinary research experience and mentoring from Roux Institute Researchers.
- Access professional development seminars on topics including networking, communication, entrepreneurial thinking, and more.
- Network with other Roux Institute researchers and partner organizations
- Explore the Greater Portland area and build community with other researchers in your cohort and entrepreneurs from our Founder Residency programs.
INTERNSHIP DURATION: June 3 – August 16, 2024
APPLICATIONS OPEN: December 15, 2024
APPLICATION DEADLINE: February 15, 2024
About The Roux Institute
The Roux Institute is designed as an engine of innovation, talent-building, and economic growth for Portland, Maine, and northern New England. We are nurturing an environment for high-impact research and innovation in computer and data science, digital engineering, advanced life sciences and medicine, and other tech fields. Partnership is what sets our education and research model apart. With leading companies at the table from day one, we are creating an agile workforce prepared to thrive in a competitive landscape powered by artificial intelligence as a force for good, and an environment for innovation in the life sciences and other high-growth sectors. Together, we envision an innovation corridor that in the coming years will stretch from Boston to Portland and beyond.
Research Internship - Causal Machine Learning in Health and Medicine
This research project explores the use of causal methods in machine learning in the domain of computer vision and medical imaging. Deep learning applications are typically driven by networks that model statistical associations, leaving them vulnerable to predictions based on shortcuts. For instance, a model designed to look for lung abnormalities based on chest X-rays might achieve high accuracies when trained on hospital data by finding evidence of surgical interventions such as chest tubes, rather than focusing on the underlying lung structure. When applied to a new set of patients with no surgical history, the model fails without access to the shortcut. Similarly, deep learning models might learn to make predictions based on sensitive characteristics, leading to bias.
Standard techniques for dealing with spurious correlations or biased predictions are to manually control for them in the training data or deep learning model. In this project, we will explore the use of causality in computer vision and medical imaging, to automatically mitigate spurious correlations. Our goal will be to design deep neural network models that automatically account for causal factors.
Under the supervision of Dr. Michael Wan, a Research Scientist at the Roux Institute working in the Institute for Experiential AI, the research intern will research relevant literature in machine learning, statistics, and medical imaging; implement causal-aware machine learning models; and write up research results for publication. Strong candidates will have proficiency in one or more of the relevant fields of study (machine learning, deep learning, computer vision, medical imaging, statistics), but certainly not all of them. Both theoretical and hands-on learning will be part of the regular responsibilities, and the project can be shifted in either direction to fit the candidate’s experience and evolving project needs. There may also be opportunities to collaborate with other Roux Institute faculty, researchers, and students, in both the health and machine learning fields.
- Research the state-of-the-art literature (papers, textbooks) on topics in machine learning, deep learning, and computer vision relevant to causal medical imaging
- Implement cutting-edge deep learning models, curate or create relevant datasets, apply the models to the datasets, analyze the results, make improvements to the models, and repeat as necessary
- Write up results for publication in machine learning conferences or journal venues
- A bachelor’s degree in a relevant computing or data discipline, such as machine learning, deep learning, computer vision, medical imaging, statistics, applied mathematics, or similar.
- Demonstrated experience with data-related programming languages such as Python, R, or a scientific computing language like MATLAB.
- Excellent communication and organizational skills
- Ability to collaborate effectively with a multidisciplinary team
- A resume and cover letter detailing your interest and qualifications for the position.
Northeastern University considers factors such as candidate work experience, education and skills when extending an offer.
Northeastern has a comprehensive benefits package for benefit eligible employees. This includes medical, vision, dental, paid time off, tuition assistance, wellness & life, retirement- as well as commuting & transportation. Visit https://hr.northeastern.edu/benefits/ for more information.
Northeastern University is an equal opportunity employer, seeking to recruit and support a broadly diverse community of faculty and staff. Northeastern values and celebrates diversity in all its forms and strives to foster an inclusive culture built on respect that affirms inter-group relations and builds cohesion.
All qualified applicants are encouraged to apply and will receive consideration for employment without regard to race, religion, color, national origin, age, sex, sexual orientation, disability status, or any other characteristic protected by applicable law.
To learn more about Northeastern University’s commitment and support of diversity and inclusion, please see www.northeastern.edu/diversity.