Potential New PhD Students
My lab is recruiting multiple PhD students for Fall 2025 at the University of Rochester. There is no application fee.
Our lab is at the forefront of deep learning, with a focus on efficient continual/lifelong machine learning, bias robust neural networks, self-supervised learning, and multi-modal large language models. We’re also proud to have a grant for using generative AI for advancing nuclear fusion research, and a new NSF award for brain-inspired models for continual learning in temporal decision making tasks. These unique research opportunities set our lab apart and offer a stimulating environment for PhD students.
As a full-time PhD student in our lab, you’ll receive a full-tuition scholarship and a stipend, both guaranteed for five years. We’re looking for applicants with strong programming skills in Python, significant deep learning experience, and a solid foundation in linear algebra and calculus. We welcome applicants from diverse academic backgrounds, including neuroscience, cognitive science, physics, math, and similar disciplines. Your training in these disciplines can greatly contribute to the empirical science of deep learning, helping you to conduct sound science, design experiments, and rigorously test hypotheses.
If you want to work in my lab, please highlight that in your statement of purpose essay. Your statement of purpose should describe why you want a PhD, your career objectives, your past research experiences, how your training has prepared you for conducting research in deep learning, and discuss at a high level the areas you want to work on. Much of the work in my lab is brain-inspired, so some familiarity with neuroscience or a willingness to learn is needed for some projects. Make sure to mention your desire to work in my lab in your essay and why you think you would be a good fit for my lab. Applicants will be reviewed based on their research and deep learning skills, especially as demonstrated by authoring papers in highly respected AI venues (CVPR, NeurIPS, ICLR, ICML, etc.). We use PyTorch in my lab. Strong alignment in our research interests is essential. I also look for strong communication abilities and grit. You should demonstrate familiarity with my recent work, where I was the papers’ senior (last) author. Except for the nuclear fusion project, the others are related to continual learning. You can find some recent preprints here. We aim to progressively train deep neural networks for large-scale language and vision problems efficiently from growing streams of (biased) large-scale datasets. I also value work experience in AI and winning AI competitions (e.g., Kaggle). Your letters of recommendation should provide supporting evidence of your skills.
In addition to your statement of purpose, The University of Rochester requires a diversity statement in your application package. You can find advice on writing one here. GRE scores are not required. If you are a US citizen, there are university-specific fellowships you may be eligible for (beyond the funding I have in my lab). You may want to call attention to your citizenship in your diversity essay. Women and underrepresented minorities in AI (Black/Hispanic) are especially encouraged to apply. Your essay should highlight any obstacles you had to overcome to achieve your educational objectives, including those related to the society you grew up in, and any activities you have done to help others overcome barriers to education.
Apply here for Fall admission before January 1. I’ll interview top-candidates in January and February.
While our normal pipeline is for students to be admitted in the Fall, exceptionally strong candidates may be considered for Spring admission.
If you have already published papers in continual learning or papers in premier journals (TPAMI, Nature, etc.) or conferences (CVPR, ICCV, NeurIPS, ICLR, ACL, AAAI, ICML, NAACL), please email me. In past years, there were 300+ applications to join my lab, although we typically only can admit 3-4 students per year. In 2024, both admitted PhD students to my lab had first-author papers in high-impact AI conferences, although in past years, this has varied considerably.
Due to the high volume of inquiries, I cannot meet with or respond to most applicants who e-mail me.