CASE STUDY

@

Pillow Lab

Powering Neuroscience Research

OVERVIEW

The Neuroscience Institute at Princeton University is home to the Pillow Lab, a computational neuroscience and statistical machine learning group developing statistical methods for studying neural systems and behavior. They collaborate closely with experimental groups to characterize the way neural populations encode, decode, and process information. At the time of writing, their research is focused on topics including approximate Bayesian inference, high-dimensional point processes, and unsupervised deep latent variable models.

  • Neuroscience and computation group led by Professor Jonathan Pillow Ph.D.
  • Research focus: neural systems and behavior

BACKGROUND

In the field of neuroscience research, there is a constant demand for flexible, powerful ways for understanding the structure in research data. Computational models aid this effort, but can come at the expense of extensive computing power and high computing costs. This was the case in the Pillow Lab, where the models required resources beyond the scope of the department. 


Researchers at the Pillow Lab previously worked on a single development server shared across the entire department, and deployed to one on-prem GPU cluster shared across the entire university. As such, experiments and training on large datasets was slow, and managing model lineage was highly inefficient. Unreliable hardware and an overworked on-prem GPU cluster inspired a search for a machine learning platform that would allow the Pillow Lab to spin-up clusters quickly and securely.

“[Truly] good science needs quite a bit of compute, and Spell helps with that.”

Daniel Greenidge, Research Assistant at the Pillow Lab

SOLUTION

The Pillow Lab implemented Spell’s cloud solution, enabling them to scale experimentation and access much-needed additional computing power. With the help of seamless onboarding and responsive support, there was no downtime or disruption of their existing workflow in the transition process, helping them save time. 


“The big advantage of Spell is that I can develop cheaply with a single instance and when I need to run a huge amount of compute I can scale up to that, run it, get the answers back at the same time it would take to train one model,” says Daniel Greenidge, a research assistant at the Pillow Lab. 


Because Spell was built by engineers for engineers, the API design and core functionality of Spell made it a true fit for their platform criteria; flexibility, efficiency, and intuitive design. 

RESULTS

Spell is making it easy for the Pillow Lab to run computationally intensive code and scale on an as-needed basis, a game changer for their research. Now, powerful compute resources are available to the group 24/7, resources that the lab wouldn’t have otherwise. This includes ongoing engagement with Spell’s machine learning Support Team. On Spell support, Daniel says, “I’m very grateful and very impressed. I have never had this level of engagement.” 


In academia, speed matters, and Spell continues to help the lab get to results, work quickly, and iterate efficiently where they could not before. Now, the Pillow Lab researchers reap the benefit of a platform that enables continuous acceleration of results and cost savings, publishing research at a much faster rate.

Streamline Machine Learning Projects with Spell

Request a Demo

Schedule an in-depth demonstration with a Spell representative to learn how Spell can help streamline and accelerate your machine learning development.

Spell is a powerful platform for building and managing machine learning projects. Spell takes care of infrastructure, making machine learning projects easier to start, faster to get results, more organized and safer than managing infrastructure on your own.


contact@spell.ml