Home

We aim to understand how large populations of neurons in the brain perform computations and represent intention. We use these insights to develop high-performance, robust, and practical assistive devices for people with disabilities and neurological disorders.

Research Highlights

Deep learning methods to uncover neural population dynamics “High-performance neural population dynamics modeling enabled by scalable computational infrastructure”. Patel*, Sedler* et al., Journal of Open Source Software 2023. [Code repository]. “Expressive architectures enhance interpretability of dynamics-based neural population models”. Sedler et al., Neurons, Behavior, Data analysis, and Theory 2023. [arXiv preprint] “A deep learning framework for inference of single-trial neural population activity from calcium imaging with sub-frame temporal resolution”. Zhu et al., Nature Neuroscience 2022. [Free access] [Tutorial and code] [Watch Feng’s Cosyne 2021 overview] “A large-scale neural network training framework for generalized estimation of single-trial population dynamics”. Keshtkaran*, Sedler* et al., Nature Methods 2022. [Free access] [Tutorial and code] [Watch Andrew’s Neuromatch 2020 talk] “Estimating muscle activation from EMG using deep learning-based dynamical systems models”. Wimalasena et al., Journal of Neural Engineering 2022. [Tutorial and code] “Deep inference of latent dynamics with spatio-temporal super resolution using selective backpropagation through time”. Zhu*, Sedler* et al., NeurIPS 2021. [Code demo] “Neural Latents Benchmark ’21: Evaluating latent variable models of neural population activity”. Pei*, Ye* et al., NeurIPS 2021 (Track on Datasets and Benchmarks). [Data, code, tutorials, evaluation APIs and more available: neurallatents.github.io] “Representation learning for neural population activity with Neural Data Transformers”. Ye & Pandarinath, Neurons, Behavior, Data analysis, and Theory 2021. [Code] [Watch Joel’s Neuromatch 2020 talk] “Enabling hyperparameter optimization in sequential autoencoders for spiking neural data”. Keshtkaran & Pandarinath, NeurIPS 2019. [Paper] [Preprint] “Inferring single-trial neural population dynamics using sequential auto-encoders”. Pandarinath et al., Nature Methods 2018. [Code (model)] [Tutorial and code (run manager)] [Free access] News & Views: “Deep learning reaches the motor system“, Nature Methods, Batista & DiCarlo Highlighted in: “How AI and neuroscience drive each other forwards“, Nature Outlook, Savage Highlighted in: F1000 Related coverage: [Emory News] [Sunday Express (UK)]

Notes to prospective trainees:

• A fair number of people of various background write about positions in the lab. I’m really, truly sorry I can’t reply to all of them. If I did reply to every email, then I wouldn’t be doing my job furthering the careers of the trainees already in my lab. I’ll try my best. (That doesn’t mean “Don’t email me!”. Just please understand, don’t get discouraged, and feel free to write again if you don’t hear back.)
• Postdocs: Exceptional postdoc candidates will always be considered.
• PhD students: All PhD students joining the lab would first need to be accepted by one of the relevant graduate programs at Emory or Georgia Tech – typically Biomedical Engineering, Neuroscience, or Machine learning, but also Electrical Engineering or Bioengineering. At Emory & GT, it is not possible to join the lab until a student is first accepted to a graduate program. Also, it is hard to project hiring far in advance, unfortunately, because it depends on funding, space on projects, and which students are the best fit in the lab. Some years I expected to hire and did not, other years I did not expect to hire, and ended up taking on multiple students. I encourage prospective students to submit their application to the program that offers the best fit. But if you are applying, and are interested in the lab, it’s super-helpful to receive a head’s up email outlining your background, interest, and which program(s) you’re applying to, so I can keep an eye out for the application!
• Masters students: In general, I don’t hire Masters students into the lab. Masters are a really short, packed period, and it’s hard to justify the amount of training/onboarding that I and other lab members have to commit. If your background is really well-aligned with what we do in the lab, and you think you can commit a really substantial amount of time, let me know. Sometimes a special projects course for a semester may help evaluate if there’s potential for a productive, longer-term research experience.
• Undergraduate students: We’ve had some great undergraduate students in the lab. Students can receive course credit, or receive funding (either through institutions or potentially lab funding). Projects range from more computer science/AI-heavy to more neuroscience/experiment-heavy. There’s room for people across this spectrum of technical expertise and backgrounds, but everyone in the lab has a genuine interest in neuroscience and neural engineering. Make sure you know the relevant deadlines for e.g. getting course credit, write to me well in advance, and stay on top of those deadlines.
• If you’re interested in the lab, definitely read through our lab manual. There’s a ton of info there, including general lab environment, policies, helpful background reading, etc.

More notes (esp. for PhD students):

• A lot of PhD applicants want to know if there will be openings in the lab the following year. I have only been here for a few years, but it’s been very hard for me to predict whether I’ll be hiring new students the following years. Some years I thought I would hire but did not, other years I thought I would not hire, yet ended up hiring multiple people. It’s a combination of factors that are very hard to predict: whether I’ll have new funding, whether I meet excellent students that are a good fit for our lab, and whether there are openings on projects that align with those students’ interests. So, many factors affect whether we’ll be hiring, and if so, who will join.
• Grad school is complex, and choosing a great mentor is critical for your career. Do your homework! A helpful starting point is Ben Barres’s article, How to Pick a Graduate Advisor.
• Here’s a page from the Shackman lab with great tips and resources on the grad school applications process.
• Since I can’t accept students directly into the lab (they have to be accepted into one of the above programs first), the best course of action is to apply to the program that’s most appropriate for your background/interests. That being said, writing to me, and letting me know which program your applying to, does help me keep a lookout for your application during admissions season. (If you’re a good candidate, it’s always helpful to make the relevant faculty aware of your application: you never know who is on the admissions committee, and those people likely don’t have time to reach out to all relevant faculty to get their opinions on applications.)

---

We are extremely grateful to the diverse funding sources that have supported or currently support our research including

• Emory Neuromodulation Technology Innovation Center (ENTICe)
• National Science Foundation (NSF NCS 1835364)
• Defense Advanced Research Projects Agency (DARPA PA-18-02-04-INI-FP-021)
• Interdisciplinary Rehabilitation Engineering Research Career Development Program through Northwestern University (NIH Eunice Kennedy Shriver NICHD K12HD073945)
• Alfred P. Sloan Foundation
• Burroughs Wellcome Fund
• Simons Foundation as part of the Simons-Emory International Consortium on Motor Control
• McCamish Parkinson’s Disease Research Program
• Emory Udall Center for Parkinson’s Research
• NIH-NINDS/OD (1DP2NS127291)
• NIH BRAIN Initiative/NIDA (1RF1DA055667)