Humans are so good in describing things in a flash. We can easily detail anything we see down to the very bits. While computers are significantly faster than humans, that simple task is extremely hard for them to understand.
On September 20th, 2016, IBM announces a multi-year collaboration with the Department of Brain & Cognitive Sciences at MIT to "advance the scientific field of machine vision, a core aspect of artificial intelligence."
The collaboration between the two is called the IBM-MIT Laboratory for Brain-inspired Multimedia Machine Comprehension’s (BM3C), and its goal is to develop cognitive ability that mimic human's.
Here, they want to create an AI that has the ability to understand and integrate inputs from multiple sources of audio and visual information. The data are then made into a computer representation of the world where the details can then be used in a variety of computer applications in industries.
The goal is to address technical challenges around both pattern and prediction method in machine vision - something that is now impossible for machines to accomplish.
Humans can easily recognize and produce detailed description of what's what, and at the same time able to predict the variety of subsequent events. BM3C wants to make computers to have that ability.
As a start, the BMC3 collaboration will bring together leading brain, cognitive, and computer scientists to conduct research in the field of unsupervised machine understanding of audio-visual streams of data. The vision of this research will lead to advances that can make computers to become more aware of their surroundings, and also to change our lives where humans and machines can work together collaboratively.
"In a world where humans and machines are working together in increasingly collaborative relationships, breakthroughs in the field of machine vision will potentially help us live healthier more productive lives," explained Guru Banavar, Chief Scientist, Cognitive Computing and VP at IBM Research. "By bringing together brain researchers and computer scientists to solve this complex technical challenge, we will advance the state-of-the-art in AI with our collaborators at MIT."
The team at MB3C will be led by Professor James DiCarlo, head of MIT's Department of Brain & Cognitive Sciences (BCS). He will be working with a team of faculty members, researchers, and graduate students from the Brain & Cognitive Sciences department and the MIT Computer Science and Artificial Intelligence Lab (CSAIL).
MIT researchers will collaborate with IBM scientists and engineers who provide technology and knowledge from IBM Watson.
Humans have been so good in identifying objects. Computers that can be so great in capturing details, still can't separate one objects from another.
For example; a camera can focus for hundreds of meters to get high-pixel resolution photos. Our eyes can't do that. But a camera and computers aren't able to track objects and differentiate a person from the background view. Humans can identify objects on the fly, we can also establish the relationships between them. This is where our vision beats computers.
BMC3 uses virtual neural networks modeled on how human's neural networks operate. Using this method, researchers have advanced computers on how they interpret the world.
The IBM-MIT partnership is trying to create the foundation of AI, and making them work to "feel" the world like humans do. Other related AI researchers include deep learning for languages, cybersecurity, decision making and so forth.