View this email in your browser
Funding a fourth year of AI-related innovation

We are pleased to announce funding for about two dozen new research projects in the coming year. A handful of our accepted proposals this time will focus on scalable real-world applications: translating computer-aided analysis of medical images to improve surgery outcomes; balancing risk and returns in a financial portfolio; and managing traffic in a hybrid system of autonomous and human-driven vehicles. This new work comes on top of the Lab-sponsored AI-related projects announced last spring to address the pandemic and its aftermath. We thank you for supporting the Lab, and wish you a healthy and productive semester ahead. 
Aude Oliva, MIT director of the MIT-IBM Watson AI Lab
David Cox, IBM director of the MIT-IBM Watson AI Lab


Shrinking deep learning's carbon footprint

To reduce the energy needed to train modern AI, researchers are experimenting with ways to make software and hardware more energy efficient, and in some cases, more like the human brain. “Humans don’t pay attention to every last detail — why should our models?” IBM's Rogerio Feris asks MIT News.
Toward an AI that recognizes abstract events

In a study at the European Conference on Computer Vision, researchers show that a deep learning model can compare and contrast a set of dynamic events on video to tease out the high-level concepts connecting them. "A model that can recognize abstract events will give more accurate, logical predictions and be more useful for decision-making,” Aude Oliva, MIT director of the Lab, tells MIT News.

Modeling circuits on the human brain

Researchers have created a new type of memory transistor, or memristor, that simulates the way that neurons in the brain store and send information. Led by MIT's Jeehwan Kim, the work could pave the way for smaller, faster computer chips that are better suited for running AI applications on hand-held devices. Read the story in MIT News and Popular Mechanics.

Separating sounds with body movements

Researchers used skeletal keypoint data to match the movements of musicians on video with the tempo of their parts, allowing listeners to isolate similar-sounding instruments. “Body keypoints provide powerful structural information,” says IBM's Chuang Gan. Read the story in MIT News and Interesting Engineering

Marshaling AI in the fight against Covid-19 

This spring the Lab announced funding for 10 AI-related projects at MIT aimed at addressing Covid-19 and its social and economic consequences. The research will target the immediate public health and economic challenges of this moment. But it could have a lasting impact on how we evaluate and respond to risk long after the pandemic has passed.

How to ensure the U.S.'s Quantum Future

"Diversity is our strength and competitive advantage," IBM Research director Dario Gil writes in this op-ed in Scientific American. "It is time to reset America’s commitment to science and to raise our level of ambition to ensure our country remains a beacon to, and the home of, the world’s best STEM talent."

In the media

Neuro-symbolic AI to give us machines with true common sense

AI systems should be able to recognize objects and reason about them, writes IBM's Katia Moskvitch in Medium. In a DARPA-funded project, MIT and IBM researchers are training computers to develop a common sense understanding of the world. 
Empowering human creativity with artificial intelligence

"Can we have AI be a co-pilot?" asks David Cox, IBM director of the Lab. "Can we have a tool that let's us build things we'd struggle to build on our own?" In an interview with ZDNet, Cox explains how machines can enhance creativity and help us achieve goals we couldn't accomplish alone.

In the virtual world

Watch: The path to building more flexible AI 

In this July roundtable discussion, experts from MIT and IBM gathered to discuss some of the challenges of developing AI systems that perform optimally in the real world. Joining the panel were MIT professors Leslie Kaelbling and Josh Tenenbaum, and IBM director of the Lab, David Cox. IDC analyst David Schubmehl moderated the event.
MIT-IBM Watson AI Lab, 75 Binney Street
Cambridge, Mass., 02142
unsubscribe | subscribe and update preferences