Research labs across the globe are doing some amazing things and pushing the envelope of what has been thought possible for data science and beyond. These research labs are among the few that are working on projects that will reshape our world in the next two generations. From autonomous vehicles, monitoring Parkinson’s Disease with a simple at-home device, to reverse engineering neural network kernels, and more, these AI research labs are at the cutting edge of data science.

So, let’s dive into these AI research labs, what they specialize in, and what they’re currently working on. 

UCLA StarAI Lab

In the heart of Los Angeles, the UCLA StarAI Lab works on the cutting edge of machine learning, statistical learning, graphic models, and the broader field of artificial intelligence. Directed by Professor Guy Van den Broeck, the lab is well known in the data science field with multiple awards and widely cited research papers.

In a recent paper from this year, StarAL Lab’s director co-authored a report detailing the new framework that unifies entropy regularization and neuro-symbolic called neuro-symbolic entropy regularization that encourages the improvement of prediction models with this combined framework.

Berkeley Artificial Intelligence Research

Berkeley Artificial Intelligence Research (BAIR) Lab boasts itself as a place where researchers interested in multi-modal deep learning, human-compatible AI, and more are able to connect the power of artificial intelligence with other scientific disciplines as well as the humanities. 

Recently, the team wrote about a concept called Reverse Engineering the Neural Tangent Kernel, in order to bring principle into architecture design, as normally deep neural networks are notoriously unprincipled. 

Brain Team at Google Research

The Brain Team is an exploratory lab founded in 2011 and is part of the Google Research family. Their goal is to rethink machine learning, which has resulted in the development of TensorFlow, and other machine learning breakthroughs. 

Currently, the team is working on a new open-source research project that looks to bridge machine learning and the creative process. Called Magenta, the project focuses on audio synthesis, music modeling, and more. 

McCormick School of Engineering – Center for Deep Learning

The center prides itself as a place where companies are able to either establish or improve their access to artificial intelligence tools through partnerships with the center. In doing so, they focus on providing technical experience and other resources by allowing members to achieve proof of concept and deployment.

Right now, the Center for Deep Learning is working on multiple projects such as Churn Prediction for CRM which looks to utilize neural networks to allow companies to capture every interaction between them and their customers. 

RiseLab at UC Berkely

Working on the idea of utilizing sensor information that everyone carries around, phones, and immersing AI, RISELab’s mission is to bring real-time applications to interact in an intelligent and secure manner. This lab works with a wide variety of topics such as software development, research, open-source software, and everything in between. 

With over forty current projects, such as AutoPandas & Dropbox alternative, Metal, RiseLab’s researchers and collaborators are working tirelessly to expand the application of AI.

MIT CSAIL’s Michael Stonebraker

Pioneering new research into computers, MIT’s CSAIL works to use emerging technology as a means to improve how people work, learn, and even enjoy their lives. Currently, the center has over sixty research groups with hundreds of projects, all seeking to find ways to make machines more intelligent, user-friendly secure, and efficient. Professor Michael Stonebraker’s project, Building a Scalable Database for Autonomous Vehicles, could help push forward a future where being hands-on with a car is an option, not a requirement. 

Another major project from the lab is an artificial intelligence breakthrough that allows in-home devices to track the progression of Parkinson’s Disease in patients by monitoring the way they breathe during sleep hours. 

PRiML at Penn State

A joint venture between Penn Engineering and The Wharton School, PRiML at Penn State seeks to bring together a large and diverse group of machine learning experts at Penn. As a forum, it hopes to engage the community in seminars, talks, and spotlight sessions that allow for a continued dialog in the world of machine learning.  

PRiML’s latest session on September 9th, saw Peking University’s Yuqing Kong speak on the topic of Eliciting & Aggregating Information with Verification. Be sure to keep an eye on future sessions from PRiML.

Stanford AI Lab

Founded in 1962, The Stanford Artificial Intelligence Laboratory (SAIL) is one of the early centers of artificial intelligence research, teaching, practice, and theory. By using a multidisciplinary and multi-faculty approach, the center promotes new discoveries and explores new ways in which humans and robots are able to interact. 

In a recent interview with Wired, Stanford’s Chelsea Finn worked to explain Moravec’s Paradox, the difficulty of observing multiple objects robots have compared to humans, to five different people; from a child to an expert. 

MURGe-Lab

Multimodal Understanding, Reasoning, and Generation for Language Lab, or MURGe-Lab, is a center at the University of North Carolina Chapel Hill that specializes in machine learning, statistical natural language processing, human-like language generation, interpretable, structured deep learning, and more.  

This year, the lab has released multiple papers for the 2022 Annual Conference of the North American Chapter of the Association of Computational Linguistics on topics like explainability, NLU, and more.

How to learn more about this data science research and AI research labs

At ODSC West 2022 coming up this November 1st-3rd, 2022, you can hear from all of these research labs and more. At the conference, researchers from leading data science and AI research labs will showcase their latest findings, explain how they use them in practical settings, and what it may mean for the field as a whole. Here are the upcoming sessions as part of the data science research track:

Artificial Intelligence Can Learn from Data. But Can It Learn to Reason? | Guy Van den Broeck, PhD | Director, Associate Professor | StarAI (Statistical and Relational Artificial Intelligence Lab), UCLA

Super Data Science Podcast with Jon Krohn: Responsible Decentralized Intelligence featuring Dawn Song | Dr. Jon Krohn | Chief Data Scientist | Nebula.io / Dawn Song, PhD | Professor | BAIR Lab, University of California – Berkeley

MLOps for Deep Learning | Diego Klabjan, PhD | Professor | Northwestern University / Yegna Jambunath | ML Ops Research Specialist | Northwestern University

AI4Cyber: An Overview of the Field and an Open-Source Virtual Machine for Research and Education | Sagar Samtani, PhD | Assistant Professor | Indiana University

Cloud Directions, MLOps, and Production Data Science | Joe Hellerstein, PhD | Jim Gray Professor of Computer Science | University of California, Berkeley

A Tale of Adversarial Attacks & Out-of-Distribution Detection Stories in the Activation Space | Celia Cintas, PhD | Research Scientist | IBM Research Africa – Nairobi

The Next Thousand Languages | Steven Bird, PhD | Professor | Charles Darwin University

Data Science Without Data Collection Using FedScale | Mosharaf Chowdhury | Associate Professor of CSE | University of Michigan

Robust and Equitable Uncertainty Estimation | Aaron Roth, PhD | Professor of Computer and Cognitive Science | University of Pennsylvania

Confidential Data Analytics and Learning for Data Scientists | Raluca Ada Popa, PhD | Associate Professor, Co-Founder | Berkeley, PreVeil and Opaque Systems

Navigating the Pitfalls of Applying Machine Learning in Practice | Jacob Schreiber | Postdoctoral Researcher | Stanford University

Denoising Diffusion-based Generative Modeling | Stefano Ermon, PhD | Assistant Professor | Stanford University