Latest in Robotics news thread

CES 2016: IBM bringing Watson's brainpower to SoftBank's Pepper robot

Being a supercomputer armed with cognitive learning powers, IBM Watson has plans to grow more powerful and maybe even omnipresent in 2016.


The machine learning wunderkind has an ambitious year ahead of itself with a bevy of new connected alliances made by possible by the wave flooding the Consumer Electronics Show this week: the Internet of Things.

The first step for introducing Watson into people's lives -- beyond what they might have seen on Jeopardy! -- is getting Watson into people's homes.

IBM has already made baby steps here through a variety of app experiments, including a collaboration last year with celebrated food magazine Bon Appétit, bringing the glossy's vast library of thousands of recipes accumulated over the year online and into a database for further culinary innovations.

But while connecting Watson to traditional home appliances -- many of which are now coming online thanks to the proliferation of the Internet of Things -- might foster awareness for machine learning capabilities faster.
 
Watson to Gain Ability to "See" with Planned $1B Acquisition of Merge Healthcare


On August 6, 2015 IBM announced that Watson will gain the ability to "see" by bringing together Watson's advanced image analytics and cognitive capabilities with data and images obtained from Merge Healthcare Incorporated's medical imaging management platform.

The vision is that these organizations could use the Watson Health Cloud to surface new insights from a consolidated, patient-centric view of current and historical images, electronic health records, data from wearable devices and other related medical data, in a HIPAA-enabled environment.
 
Quote
This movie shows 20 different people drawing a novel character (left) and the algorithm predicting how those images were drawn (right). Credit: Brenden Lake
According to Science, researchers at MIT, NYU and UofT have demonstrated the capacity of an algorithm to both achieve human-level concept learning and to consistently pass a visual “Turing-test”.
In the words of one of the researchers, Dr. Joshua Tannenbaum, the authors believe that they have created “for the first time… a machine system that can learn a large class of visual concepts in ways that are hard to distinguish from human learners.”
The long-term goal of the authors’ work is to narrow the gap between machine and human learning. An important aspect of this gap is the capacity of human beings to be able to learn concepts with relatively few examples, or even from a single example. For humans, learning to mimic a hand gesture or a written character is something that can be achieved with very few repetitions.
Current machine learning approaches, on the other hand, require algorithms to be trained using tens of thousands of examples. In 2006, for example, the aforementioned authors conducted a study in which they were able to teach an algorithm the structure of 10 handwritten characters, but that process required 60,000 examples.
Jump ahead to today where, using an approach they have called “Bayesian Program Learning” (BPL), the authors have demonstrated the capacity of BPL to replicate human level performance after only a single exposure to visual content.​
 
Deep Learning Helps Robot Learn to Walk the Way Humans Do
Darwin the robot wobbles when he walks. And sometimes he falls. But unlike most robots, he learns from his mistakes — just like people do — and adjusts his technique on the fly.
His baby steps could lead to a new generation of autonomous robots that adapt to changing environments and new situations without a human reprogramming them. These robots could tackle dangerous tasks such as handling rescue efforts or cleaning up disaster areas. Or, they could become assistants that help out around the house or ferry a package across town.
“An autonomous robot would be able to take a high-level goal and figure out how to achieve it,” said Igor Mordatch, a postdoctoral fellow at the University of California, Berkeley, who is leading the Darwin research project. “That would be very, very powerful.”​


Housekeeping ATLAS-style: Tedious but each minute counts

Sweeping up, tidying up, pushing the handle of a cleaner back and forth, kneeling for a stray paper and depositing it into a wastebasket, YAWN.
 
Last edited:

New Topics

Forum List

Back
Top